markdown
stringlengths
0
1.02M
code
stringlengths
0
832k
output
stringlengths
0
1.02M
license
stringlengths
3
36
path
stringlengths
6
265
repo_name
stringlengths
6
127
Basic Functions![](http://)**Hint** Be sure to `return` values from your function definitions. The assert statements will call your function(s) for you.
# Run this cell in order to generate some numbers to use in our functions after this. import random positive_even_number = random.randrange(2, 101, 2) negative_even_number = random.randrange(-100, -1, 2) positive_odd_number = random.randrange(1, 100, 2) negative_odd_number = random.randrange(-101, 0, 2) print("We...
Exercise 42 is correct.
MIT
101_exercises.ipynb
barbmarques/python-exercises
Functions working with stringsIf you need some guidance working with the next few problems, recommend reading through [this example code](https://gist.github.com/ryanorsinger/f758599c886549e7615ec43488ae514c)
# Exercise 43 # Write a function definition named is_vowel that takes in value and returns True if the value is a, e, i, o, u in upper or lower case. def is_vowel(x): return x.lower() in "aeiou" assert is_vowel("a") == True assert is_vowel("U") == True assert is_vowel("banana") == False assert is_vowel("Q") == Fals...
Exercise 49 is correct.
MIT
101_exercises.ipynb
barbmarques/python-exercises
Accessing List Elements
# Exercise 50 # Write a function definition named first that takes in sequence and returns the first value of that sequence. def first(x): return x[0] assert first("ubuntu") == "u" assert first([1, 2, 3]) == 1 assert first(["python", "is", "awesome"]) == "python" print("Exercise 50 is correct.") # Exercise 51 # Wri...
Exercise 59 is correct.
MIT
101_exercises.ipynb
barbmarques/python-exercises
Functions to describe data
# Exercise 60 # Write a function definition named sum_all that takes in sequence of numbers and returns all the numbers added together. def sum_all(numbers): return sum(numbers) assert sum_all([1, 2, 3, 4]) == 10 assert sum_all([3, 3, 3]) == 9 assert sum_all([0, 5, 6]) == 11 print("Exercise 60 is correct.") # Exe...
Exercise 64 is correct.
MIT
101_exercises.ipynb
barbmarques/python-exercises
Applying functions to lists
# Run this cell in order to use the following list of numbers for the next exercises numbers = [-5, -4, -3, -2, -1, 1, 2, 3, 4, 5] # Exercise 65 # Write a function definition named get_highest_number that takes in sequence of numbers and returns the largest number. def get_highest_number(x): return max(x) assert g...
Exercise 82 is correct.
MIT
101_exercises.ipynb
barbmarques/python-exercises
Working with sets**Hint** Take a look at the `set` function in Python, the `set` data type, and built-in `set` methods.
# Example set function usage print(set("kiwi")) print(set([1, 2, 2, 3, 3, 3, 4, 4, 4, 4])) # Exercise 83 # Write a function definition named get_unique_values that takes in a list and returns a set with only the unique values from that list. def get_unique_values(x): return set(x) assert get_unique_values(["ant", "...
Exercise 86 is correct.
MIT
101_exercises.ipynb
barbmarques/python-exercises
Working with Dictionaries
# Run this cell in order to have these two dictionary variables defined. tukey_paper = { "title": "The Future of Data Analysis", "author": "John W. Tukey", "link": "https://projecteuclid.org/euclid.aoms/1177704711", "year_published": 1962 } thomas_paper = { "title": "A mathematical model of glutath...
Exercise 90 is complete.
MIT
101_exercises.ipynb
barbmarques/python-exercises
Working with Lists of Dictionaries**Hint** If you need an example of lists of dictionaries, see [https://gist.github.com/ryanorsinger/fce8154028a924c1073eac24c7c3f409](https://gist.github.com/ryanorsinger/fce8154028a924c1073eac24c7c3f409)
# Run this cell in order to have some setup data for the next exercises books = [ { "title": "Genetic Algorithms and Machine Learning for Programmers", "price": 36.99, "author": "Frances Buontempo" }, { "title": "The Visual Display of Quantitative Information", "price...
_____no_output_____
MIT
101_exercises.ipynb
barbmarques/python-exercises
Data Analysis for Inverse Observation Data Assimilation of Kolmogorov FlowThis notebook analyzes the paper's data and reproduces the plots.
import os os.environ["CUDA_VISIBLE_DEVICES"]="0" # system integration faster on GPU import warnings warnings.filterwarnings('ignore') from functools import partial import numpy as np import scipy import jax import jax.numpy as jnp from jax import random, jit import argparse from datetime import datetime import xarray a...
_____no_output_____
Apache-2.0
Analysis_KolmogorovFlow.ipynb
googleinterns/invobs-data-assimilation
Copy data from Google cloudThis requires [gsutil](https://cloud.google.com/storage/docs/gsutil).
!gsutil cp -r gs://gresearch/jax-cfd/projects/invobs-data-assimilation/invobs-da-results/ /tmp
Copying gs://gresearch/jax-cfd/projects/invobs-data-assimilation/invobs-da-results/kolmogorov_baselineinit_hybridopt.nc... Copying gs://gresearch/jax-cfd/projects/invobs-data-assimilation/invobs-da-results/kolmogorov_baselineinit_obsopt.nc... Copying gs://gresearch/jax-cfd/projects/invobs-data-assimilation/invobs-da-re...
Apache-2.0
Analysis_KolmogorovFlow.ipynb
googleinterns/invobs-data-assimilation
Load data
path = '/tmp/invobs-da-results' filenames = [ 'kolmogorov_baselineinit_obsopt.nc', 'kolmogorov_baselineinit_hybridopt.nc', 'kolmogorov_invobsinit_obsopt.nc', 'kolmogorov_invobsinit_hybridopt.nc', ] retained_variables = [ 'f_vals', 'eval_vals', 'X0_ground_truth', 'X0_opt', 'X0_init',...
_____no_output_____
Apache-2.0
Analysis_KolmogorovFlow.ipynb
googleinterns/invobs-data-assimilation
Instantiate dynamical system
kolmogorov_flow = KolmogorovFlow( grid_size=ds.attrs['grid_size'], num_inner_steps=ds.attrs['num_inner_steps'], viscosity=ds.attrs['viscosity'], observe_every=ds.attrs['observe_every'], wavenumber=ds.attrs['peak_wavenumber'], )
_____no_output_____
Apache-2.0
Analysis_KolmogorovFlow.ipynb
googleinterns/invobs-data-assimilation
Data assimilation initialization samplesComparison of initialization schemes.
da_init = xr.concat( [ ds['X0_init'].sel(init=['invobs', 'baseline']), ds['X0_ground_truth'].sel(init='baseline'), ], dim='init', ) \ .assign_coords(init=['invobs', 'baseline', 'ground_truth']) \ .sel(opt_space='observation') vort_init = compute_vorticity(da_init, kolmogorov_flow.grid) #...
_____no_output_____
Apache-2.0
Analysis_KolmogorovFlow.ipynb
googleinterns/invobs-data-assimilation
Optimization curvesPlot value of observation space objective function during optimization normalized by the first-step value of the observation space objective function.
sns.set(font_scale=1.5) sns.set_style('white') to_plot = ds['eval_vals'].sel(n=[6, 9]) to_plot_relative_mean = ( to_plot / to_plot.sel(opt_step=0, opt_space='observation') ) to_plot_relative_mean = to_plot_relative_mean.sel( init=['invobs', 'baseline'], opt_space=['hybrid', 'observation'], ) df_opt_curves...
_____no_output_____
Apache-2.0
Analysis_KolmogorovFlow.ipynb
googleinterns/invobs-data-assimilation
Forecast quality
X0_da = ds[['X0_ground_truth', 'X0_init', 'X0_opt']].to_array('data_type') \ .assign_coords({'data_type': ['gt', 'init', 'opt']}) X_da = integrate_kolmogorov_xr(kolmogorov_flow, X0_da, 20) vorticity = compute_vorticity(X_da, kolmogorov_flow.grid) relative_scale = 14533 # average L1 norm over independent samples l1_erro...
_____no_output_____
Apache-2.0
Analysis_KolmogorovFlow.ipynb
googleinterns/invobs-data-assimilation
Summary statsCompare forecast performance on the first forecast state relative to baseline init and optimization method.
summary_stats = l1_error.sel(data_type='opt', t=11).mean(dim='n') / l1_error.sel(data_type='opt', t=11, init='baseline', opt_space='observation').mean(dim='n') print( summary_stats.sel(opt_space='observation', init='baseline').values, summary_stats.sel(opt_space='hybrid', init='baseline').values, summary_st...
1.0 0.8789517 0.19961545 0.22848208
Apache-2.0
Analysis_KolmogorovFlow.ipynb
googleinterns/invobs-data-assimilation
Significance test between trajectoriesPerform a Z-test to evaluate significance level between optimization methods for the two initialization schemes. Inverse observation initialization
time_step = 11 # beginning of forecast window num_samples = l1_error.sizes['n'] l1_error_inv = l1_error.sel(init='invobs', data_type='opt') diff_l1_error = ( l1_error_inv.sel(opt_space='observation') - l1_error_inv.sel(opt_space='hybrid') ) m = diff_l1_error.sel(t=time_step).mean(dim='n') s = diff_l1_error.sel(t...
Z-value -5.658017083581407 p-value 7.656594807321588e-09
Apache-2.0
Analysis_KolmogorovFlow.ipynb
googleinterns/invobs-data-assimilation
Baseline initialization
time_step = 11 # beginning of forecast window num_samples = l1_error.sizes['n'] l1_error_inv = l1_error.sel(init='baseline', data_type='opt') diff_l1_error = ( l1_error_inv.sel(opt_space='observation') - l1_error_inv.sel(opt_space='hybrid') ) m = diff_l1_error.sel(t=time_step).mean(dim='n') s = diff_l1_error.sel...
Z-value 6.508546551414507 p-value 3.7940697809822585e-11
Apache-2.0
Analysis_KolmogorovFlow.ipynb
googleinterns/invobs-data-assimilation
Assimilated trajectories
gt = vorticity.sel(data_type='gt', opt_space='observation', init='baseline') baseline = vorticity.sel( data_type='opt', opt_space='observation', init='baseline', ) invobs = vorticity.sel(data_type='opt', opt_space='hybrid', init='invobs') forecast_comparison = ( xr.concat([invobs, baseline, gt], dim=...
_____no_output_____
Apache-2.0
Analysis_KolmogorovFlow.ipynb
googleinterns/invobs-data-assimilation
Summary figure
sns.set(font_scale=3) plt.rc('font', **{'family': 'Times New Roman'}) g = forecast_comparison.isel(n=1, t=11) \ .plot.imshow( x='x', y='y', col='da_method', size=5, add_colorbar=False, cmap=sns.cm.icefire, vmin=-8, vmax=8, ) col_labels = ['ground truth', 'proposed', 'baseline'] ...
_____no_output_____
Apache-2.0
Analysis_KolmogorovFlow.ipynb
googleinterns/invobs-data-assimilation
Индивидуальное задание: Изобразить эмблему Бэтмена для его призыва к спасению города от противников бабы Нины Задача: Существуют хейтеры бабы Нины и перед Бэтменом была поставлена задача спасти её от них Решение: V = 1 Бэтмен, p хейтеры = 100 штук P = pV P = 100 * 1 = 100 штук Q = V : P Q = 1 : 100= 0,01 H Ответ: Q =...
%load_ext autoreload %autoreload 2 import jupyter_lesson %matplotlib inline jupyter_lesson.plot_batman() import matplotlib.path as mpath import matplotlib.patches as mpatches import matplotlib.pyplot as plt fig, ax = plt.subplots() Path = mpath.Path path_data = [ (Path.MOVETO, (1.58, -2.57)), (Path.CURVE4,...
_____no_output_____
MIT
ind.ipynb
A1zak/cross6
Definitions of Python:1. Primitive types, basic operations2. Composed types: lists, tuples, dictionaries3. Everything is an object4. Control structures: blocks, branching, loops Primitive typesThe basic types build into Python include:* `int`: variable length integers,* `float`: double precision floating point number...
##### -1234567890 # an integer 2.0 # a floating point number 6.02e23 # a floating point number with scientific notation complex(1, 5) # a complex True or False # the two possible boolean values 'This is a string' "It's another string" print("""Triple quotes (also with '''), allow strings to break ove...
_____no_output_____
CC-BY-4.0
python/python/1_Definitions.ipynb
t20100/silx-training
Primary operations| Symbol | Task Performed ||----|---|| + | Addition || - | Subtraction || * | Multiplication || / | Floating point division || // | Floor division || % | Modulus or rest || ** or pow(a, b) | Power || abs(a) | absolute value || round(a) | Banker's rounding |Some examples:
#divisions print(3 / 2, 3 // 2) print(3. / 2, 3. // 2) #operation with complex print((5. + 4.0j - 3.5) * 2.1)
_____no_output_____
CC-BY-4.0
python/python/1_Definitions.ipynb
t20100/silx-training
Relational Operators | Symbol | Task Performed ||---|---|| == | True if it is equal || != | True if not equal to || < | less than || > | greater than || <= | less than or equal to || >= | greater than or equal to ||&nbsp;|| not | negate a `bool` value || is | True if both are the same || and | True if both are True || ...
# grouping comparison print(1 >= 0.5 and (2 > 3 or 5 % 2 == 1))
_____no_output_____
CC-BY-4.0
python/python/1_Definitions.ipynb
t20100/silx-training
Strings* Basic operations on strings: `+` or `*`* Formating using `%s` or `.format`* Slicing using the [start: stop: step]
s = "a is equal to" a = 5 print(s+str(a)) print(s, a) #/!\ s+a # String muliplied by int works ! "*--*" * 5
_____no_output_____
CC-BY-4.0
python/python/1_Definitions.ipynb
t20100/silx-training
String formating
# %s will convert anything to an int. %i, %d, %f works like in C or spec. print("a is equal to %s" % a) print("%s %05i" % (s,a)) #new style formating '{2} {1} {2} {0}'.format('a','b','c') '{0:.2} {1:.1%}'.format(0.3698, 9.6/100)
_____no_output_____
CC-BY-4.0
python/python/1_Definitions.ipynb
t20100/silx-training
String access and slicing* Access a single element with `[index]`* Extract a sub part of a string using `[start:stop:step]` → **slicing**.* Works the same on lists, tuples, arrays, ...
letters = "abcdefghijklmnopqrstuvwxyz" print(letters) len(letters) # remind you that python is 0-based indexing print(letters[0], letters[4], letters[25]) #/!\ letters[56] #slicing from beginging letters[0:4] #from the end letters[-5:] # two by two using a stepsize of 2 letters[::2] # inverted using a stepsize of -1: l...
_____no_output_____
CC-BY-4.0
python/python/1_Definitions.ipynb
t20100/silx-training
Useful string methods```pythonmy_str = 'toto'```* `len(my_str)`: returns the length of the string* `my_str.find('to')`, `my_str.index('to')`: returns the starting index. Find returns ``-1`` if not found, index fails.* `my_str.replace(str1, str2)`: replaces `str1` with `str2` in string* `my_str.split()` splits the stri...
print([1, 2, 3]) digits = list(range(10)) print(digits) print(list(letters)) print(digits + list(letters)) #but digits + letters # lists can contain anything a = ['my string', True, 5+7] print(len(a)) a.append(3.141592) print(a, len(a)) list(range(5, 12, 2))
_____no_output_____
CC-BY-4.0
python/python/1_Definitions.ipynb
t20100/silx-training
Useful methods for lists```pythonlst = [1, 2, 3, 'a', 'b', 'c']```* `lst.append(a)`: adds *a* at the end of lst* `lst.insert(idx, a)`: inserts one element at a given index* `lst.index(a)`: finds first index containing a value* `lst.count(a)`: counts the number of occurences of *a* in the list * `lst.pop(idx)`: removes...
lst = ['eggs', 'sausages'] print(len(lst)) lst.append("spam") print(lst, len(lst)) lst.insert(0, "spam") print(lst) print(lst.index("spam"), lst.index("sausages")) #but: lst.index(5) lst.count("spam") print(lst.pop(), lst.pop(2)) # lists are mutable: print(lst) lst[0] = 1 print(lst) lst.remove("eggs") print(lst) # but ...
_____no_output_____
CC-BY-4.0
python/python/1_Definitions.ipynb
t20100/silx-training
Tuple* Tuples are defined by the `tuple(iter)` or by a ``,`` separated list in parenthesis ``()`` * Tuples are like lists, but not mutable !
mytuple = ('spam', 'eggs', 5, 3.141592, 'sausages') print(mytuple[0], mytuple[-1]) # /!\ tuples are not mutable mytuple[3] = "ham" # Single element tuple: mind the comma t = 5, print(t)
_____no_output_____
CC-BY-4.0
python/python/1_Definitions.ipynb
t20100/silx-training
List comprehension and generatorsVery *pythonic* and convenient way of creating lists or tuples from an iterator: ` [ f(i) for i in iterable if condition(i) ]`The content of the `[]` is called a *generator*.A *generator* generates elements on demand, which is *fast* and *low-memory usage*. It is the base of the asynch...
[2*x+1 for x in range(5)] tuple(x**(1/2.) for x in range(5)) (x**(1/2.) for x in range(5)) [x for x in range(10) if x**3 - 15*x**2 + 71*x == 105] print([l.upper() for l in letters])
_____no_output_____
CC-BY-4.0
python/python/1_Definitions.ipynb
t20100/silx-training
Mapping Types: DictionariesDictionaries associate a key to a value using curly braces `{key1: value1, key2:value2}`: * Keys must be *hashable*, i.e. any object that is unmutable, also known as *hash table* in other languages* Dictionaries were not ordered before Python 3.7 (`OrderedDict` are)
help(dict) periodic_table = { "H": 1, "He": 2, "Li": 3, "Be": 4, "B": 5, "C": 6, "N": 7, "O": 8, "F": 9, "Ne": 10, } print(periodic_table) print(periodic_table['He']) print(periodic_table.keys()) print(periodic_table.values()) #search for a key in a dict: 'F' in periodic_tab...
_____no_output_____
CC-BY-4.0
python/python/1_Definitions.ipynb
t20100/silx-training
In Python, everything is object* In Python everything is object (inherits from ``object``)* Names are just labels, references, attached to an object* Memory is freed when the number of references drops to 0- `dir(obj)`: lists the attributes of an object- `help(obj)`: prints the help of the object- `type(obj)`: gets th...
a = object() print(dir(a)) print(type(True), type(a), id(a)) b = 5 c = 5 print(id(b), id(c), id(b) == id(c), b is c) # == vs `is` a, b = 5, 5.0 print(a == b) print(type(a), type(b)) # int are unique print(a, bin(a), a is 0b0101) print(a == 5.0, a is 5.0) print(1 is None, None is None)
_____no_output_____
CC-BY-4.0
python/python/1_Definitions.ipynb
t20100/silx-training
Warning: in Python, everything is object ...
list1 = [3, 2, 1] print("list1=", list1) list2 = list1 list2[1] = 10 print("list2=", list2) # Did you expect ? print("list1= ", list1)
_____no_output_____
CC-BY-4.0
python/python/1_Definitions.ipynb
t20100/silx-training
![a=1](img/Python_memory_1.png "a=1") ![list1](img/Python_memory_2.png "list11") ![list1](img/Python_memory_3.png "list2") ![mutate](img/Python_memory_4.png "Mutate list2") ![a=1](img/Python_memory_5.png "a=1") ![copy](img/Python_memory_6.png "copy")
print("indeed: id(list1)=", id(list1)," and id(list2):", id(list2), "so list2 is list1:", list2 is list1) #How to avoid this: make copies of mutable objects: list1 = [3, 2, 1] print("list1=", list1) list3 = list1 [:] # copy the content ! list3[1] = 10 print("As expected: list3= ", list3) print("And now: list1= ", l...
_____no_output_____
CC-BY-4.0
python/python/1_Definitions.ipynb
t20100/silx-training
**Warning:** This is very error prone when manipulating **any** mutable objects.
# Generic solution: use the copy module import copy list3 = copy.copy(list1) # same, more explicit print(id(list1) == id(list3))
_____no_output_____
CC-BY-4.0
python/python/1_Definitions.ipynb
t20100/silx-training
Control structures: blocks Code structurePython uses a colon `:` at the end of the line and 4 white-spaces indentationto establish code block structure.Many other programming languages use braces { }, not python. ``` Block 1 ... Header making new block: Block 2 ... Header making new block...
a = -1 b = 2 c = 1 q2 = b * b - 4.0 * a * c print("Determinant is ", q2) import math if q2 < 0: print("No real solution") elif q2 > 0: x1 = (-b + math.sqrt(q2)) / (2.0 * a) x2 = (-b - math.sqrt(q2)) / (2.0 * a) print("Two solutions %.2f and %.2f" % (x1, x2)) else: x = -b / (2.0 * a) print("One s...
_____no_output_____
CC-BY-4.0
python/python/1_Definitions.ipynb
t20100/silx-training
For loop- iterate over a sequence (list, tuple, char in string, keys in dict, any iterator)- no indexes, uses directly the object in the sequence- when the index is really needed, use `enumerate`- One can use multiple sequences in parallel using `zip`
ingredients = ["spam", "eggs", "ham", "spam", "sausages"] for food in ingredients: print("I like %s" % food) for idx, food in enumerate(ingredients[-1::-1]): print("%s is number %d in my top 5 of foods" % (food, len(ingredients)- idx)) subjects = ["Roses", "Violets", "Sugar"] verbs = ["are", "are", "is"] adjec...
_____no_output_____
CC-BY-4.0
python/python/1_Definitions.ipynb
t20100/silx-training
While loop- Iterate while a condition is fulfilled- Make sure the condition becomes unfulfilled, else it could result in infinite loops ...
a, b = 175, 3650 stop = False possible_divisor = max(a, b) // 2 while possible_divisor >= 1 and not stop: if a % possible_divisor == 0 and b % possible_divisor == 0: print("Found greatest common divisor: %d" % possible_divisor) stop = True possible_divisor = possible_divisor - 1 while True: ...
_____no_output_____
CC-BY-4.0
python/python/1_Definitions.ipynb
t20100/silx-training
Useful commands in loops- `continue`: go directly to the next iteration of the most inner loop- `break`: quit the most inner loop- `pass`: a block cannot be empty; ``pass`` is a command that does nothing- `else`: block executed after the normal exit of the loop.
for i in range(10): if not i % 7 == 0: print("%d is *not* a multiple of 7" % i) continue print("%d is a multiple of 7" % i) n = 112 # divide n by 2 until this does no longer return an integer while True: if n % 2 != 0: print("%d is not a multiple of 2" % n) break print("%...
_____no_output_____
CC-BY-4.0
python/python/1_Definitions.ipynb
t20100/silx-training
Exercise: Fibonacci series- Fibonacci: - Each element is the sum of the previous two elements - The first two elements are 0 and 1- Calculate all elements in this series up to 1000, put them in a list, then print the list.``[0, 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144, 233, 377, 610, 987]``prepend the cell with...
# One possible solution res = [0, 1] next_el = 1 while next_el < 1000: res.append(next_el) next_el = res[-2] + res[-1] print(res)
_____no_output_____
CC-BY-4.0
python/python/1_Definitions.ipynb
t20100/silx-training
RNNs for Timeseries Analysis Timeseries Bruno Gonçalves www.data4sci.com @bgoncalves, @data4sci
import pandas as pd from pandas.plotting import autocorrelation_plot import numpy as np np.random.seed(123) from sklearn.linear_model import LinearRegression import matplotlib.pyplot as plt import watermark %load_ext watermark %matplotlib inline %watermark -n -v -m -p numpy,matplotlib,pandas,sklearn
Wed Sep 25 2019 CPython 3.7.3 IPython 6.2.1 numpy 1.16.2 matplotlib 3.1.0 pandas 0.24.2 sklearn 0.20.3 compiler : Clang 4.0.1 (tags/RELEASE_401/final) system : Darwin release : 18.7.0 machine : x86_64 processor : i386 CPU cores : 8 interpreter: 64bit
MIT
Timeseries.ipynb
k-rajmani2k/RNN
Load the datasetGDP data from the Federal Reserve Bank [website](https://fred.stlouisfed.org/series/GDP)
series = pd.read_csv('data/GDP.csv', header=0, parse_dates=[0], index_col=0) plt.plot(series) plt.xlabel('Date') plt.ylabel('GDP (B$)'); plt.gcf().set_size_inches(11,8) series['GDP'].pct_change().plot() plt.gca().plot([series.index.min(), series.index.max()], [0, 0], 'r-') plt.xlabel('Date') plt.ylabel('GDP QoQ Growth'...
_____no_output_____
MIT
Timeseries.ipynb
k-rajmani2k/RNN
Autocorrelation function
autocorrelation_plot(series, label='DJIA') autocorrelation_plot(series.pct_change().dropna(), ax=plt.gca(), label='DoD') plt.gcf().set_size_inches(11,8) values = series.pct_change().dropna().values.reshape(-1, 1) X = values[:-1] y = values[1:] plt.plot(X.flatten(), y, '*') plt.xlabel('x_t') plt.ylabel('x_t+1') lm = Lin...
_____no_output_____
MIT
Timeseries.ipynb
k-rajmani2k/RNN
Fit comparison
plt.plot(series.index[2:], series.values[2:], ) plt.plot(series.index[2:], (1+y_pred).cumprod()*series.values[0]) plt.xlabel('Date') plt.ylabel('GDP (B$)') plt.gcf().set_size_inches(11, 8)
_____no_output_____
MIT
Timeseries.ipynb
k-rajmani2k/RNN
Now without looking into the future
n_points = len(series) train_points = int(2/3*n_points)+1 X_train = X[:train_points] y_train = y[:train_points] X_test = X[train_points:] y_test = y[train_points:] lm.fit(X_train, y_train) y_train_pred = lm.predict(X_train) y_test_pred = lm.predict(X_test) plt.plot(series.index[:train_points], y_train, label='data') pl...
_____no_output_____
MIT
Timeseries.ipynb
k-rajmani2k/RNN
Comparison plot
plt.plot(series.index[:train_points], (1+y_train).cumprod()*series.values[0], label='train') plt.plot(series.index[train_points+2:], (1+y_test).cumprod()*series.values[train_points], label='test') plt.plot(series.index[:train_points], (1+y_train_pred).cumprod()*series.values[0], label='train_pred') plt.plot(series.inde...
_____no_output_____
MIT
Timeseries.ipynb
k-rajmani2k/RNN
Parallel simulations using mpi4py
import os,sys Nthread = 1 os.environ["OMP_NUM_THREADS"] = str(Nthread) # export OMP_NUM_THREADS=1 os.environ["OPENBLAS_NUM_THREADS"] = str(Nthread) # export OPENBLAS_NUM_THREADS=1 os.environ["MKL_NUM_THREADS"] = str(Nthread) # export MKL_NUM_THREADS=1 os.environ["VECLIB_MAXIMUM_THREADS"] = str(Nthread) # export VECLIB_...
_____no_output_____
MIT
notebooks/parallel_simulations.ipynb
zhaonat/py-maxwell-fd3d
Playing TextWorld generated games with OpenAI GymThis tutorial shows how to play a text-based adventure game **generated by TextWorld** using OpenAI's Gym API. Generate a new TextWorld game
!tw-make custom --world-size 2 --quest-length 3 --nb-objects 10 --output tw_games/game.ulx -f -v --seed 1234
Global seed: 1234 Game generated: tw_games/game.ulx Welcome to another fast paced game of TextWorld! Here is your task for today. First, it would be great if you could try to go east. With that accomplished, take the shoe that's in the attic. With the shoe, place the shoe on the shelf. Alright, thanks!
MIT
notebooks/Playing TextWorld generated games with OpenAI Gym.ipynb
zhaozj89/TextWorld
Register the game with GymIn order to call to `gym.make`, we need to create a valid `env_id` for our game.
import textworld.gym env_id = textworld.gym.register_game('tw_games/game.ulx')
_____no_output_____
MIT
notebooks/Playing TextWorld generated games with OpenAI Gym.ipynb
zhaozj89/TextWorld
Make the gym environmentWith our `env_id` we are ready to use gym to start the new environment.
import gym env = gym.make(env_id)
_____no_output_____
MIT
notebooks/Playing TextWorld generated games with OpenAI Gym.ipynb
zhaozj89/TextWorld
Start the gameLike for other Gym environments, we start a new game by calling the `env.reset` method. It returns the initial observation text string as well as a dictionary for additional informations (more on that later).
obs, infos = env.reset() print(obs)
________ ________ __ __ ________ | \| \| \ | \| \ \$$$$$$$$| $$$$$$$$| $$ | $$ \$$$$$$$$ | $$ | $$__ \$$\/ $$ | $$ | $$ | $$ \ >$$ $$ ...
MIT
notebooks/Playing TextWorld generated games with OpenAI Gym.ipynb
zhaozj89/TextWorld
Interact with the gameThe `env.step` method is used to send text command to the game. This method returns the observation for the new state, the current game score, whether the game is done, and dictionary for additional informations (more on that later).
obs, score, done, infos = env.step("open box") print(obs)
You have to unlock the Canadian style box with the Canadian style key first.
MIT
notebooks/Playing TextWorld generated games with OpenAI Gym.ipynb
zhaozj89/TextWorld
Make a simple play loopWe now have everything we need in order to interactively play a text-based game.
try: done = False obs, _ = env.reset() print(obs) nb_moves = 0 while not done: command = input("> ") obs, score, done, _ = env.step(command) print(obs) nb_moves += 1 except KeyboardInterrupt: pass # Press the stop button in the toolbar to quit the game. print("...
________ ________ __ __ ________ | \| \| \ | \| \ \$$$$$$$$| $$$$$$$$| $$ | $$ \$$$$$$$$ | $$ | $$__ \$$\/ $$ | $$ | $$ | $$ \ >$$ $$ ...
MIT
notebooks/Playing TextWorld generated games with OpenAI Gym.ipynb
zhaozj89/TextWorld
Request additional information_*Only available for games generated with TextWorld._To ease the learning process of AI agents, TextWorld offers control over what information should be available alongside the game's narrative (i.e. the observation).Let's request the list of __admissible__ commands (i.e. commands that ar...
import textworld request_infos = textworld.EnvInfos(admissible_commands=True, entities=True) # Requesting additional information should be done when registering the game. env_id = textworld.gym.register_game('tw_games/game.ulx', request_infos) env = gym.make(env_id) obs, infos = env.reset() print("Entities: {}".forma...
Entities: ['safe', 'Canadian style box', 'shoe', 'cane', 'shelf', 'workbench', 'board', 'bowl', 'Canadian style key', 'book', 'fork', 'pair of pants', 'north', 'south', 'east', 'west'] Admissible commands: drop Canadian style key drop book drop fork examine Canadian style box examine Canadian style key exam...
MIT
notebooks/Playing TextWorld generated games with OpenAI Gym.ipynb
zhaozj89/TextWorld
* Problem 2 - Even Fibonacci numbers Each new term in the Fibonacci sequence is generated by adding the previous two terms. By starting with 1 and 2, the first 10 terms will be:1, 2, 3, 5, 8, 13, 21, 34, 55, 89, ... By considering the terms in the Fibonacci sequence whose values do not exceed four million, f...
if __name__ == "__main__": assert fibonacci(10) == 89 """Let's start with the 10th fibonacci term""" def fibonacci(n): return 89 if __name__ == "__main__": assert fibonacci(10) == 89 #assert fibonacci(9) == 55 def fibonacci(n): if n == 9: return 55 return 89 if __name__ == "__main__": ...
_____no_output_____
MIT
2-Even Fibonacci numbers(right).ipynb
vss-py/TDD-ProjectEuler
But if I do this way, I probabilly having a problem to apply the fibonacci algorithm,so, let's begin with the ascending order
def fibonacci(n): return 1 if __name__ == '__main__': assert fibonacci(1) == 1 def fibonacci(n): return 1 if __name__ == '__main__': assert fibonacci(1) == 1 assert fibonacci(2) == 2 def fibonacci(n): if n == 2: return 2 return 1 if __name__ == '__main__': assert fibonacci(1) ...
_____no_output_____
MIT
2-Even Fibonacci numbers(right).ipynb
vss-py/TDD-ProjectEuler
What the code is doing is: n = (5 - 2) + (5 - 1).Therefore we must to apply a repetition.
def fibonacci(n): a = 1 b = 2 c = 0 i = 0 if n > 2: while i < n: c = a + b a = b b = c i = i + 1 return c if n == 2: return n return 1 if __name__ == '__main__': assert fibonacci(1) == 1 assert fibonacci(2) == 2 ...
_____no_output_____
MIT
2-Even Fibonacci numbers(right).ipynb
vss-py/TDD-ProjectEuler
* Now, I must to find the list (the second step)As I am using the TDD, I can do this without a problem.
def fibonacci(n): a = 1 b = 2 c = 0 i = 2 if n > 2: while i < n: c = a + b a = b b = c i = i + 1 return c if n == 2: return n return 1 if __name__ == '__main__': assert fibonacci(1) == 1 assert fibonacci(2) == 2 ...
_____no_output_____
MIT
2-Even Fibonacci numbers(right).ipynb
vss-py/TDD-ProjectEuler
The repetition of the first term occurs because I don't set what the function should to do if n < 2.
def fibonacci_list(l): lista = [] i = 1 while i < l: a = fibonacci(i) lista.append(a) i = i + 1 return lista def fibonacci(n): a = 1 b = 2 c = 0 i = 2 if n > 2: while i < n: c = a + b a = b b = c i = i + ...
_____no_output_____
MIT
2-Even Fibonacci numbers(right).ipynb
vss-py/TDD-ProjectEuler
Now, I must to apply the condition that I miss. To append in the list only the even terms.
def fibonacci_list(l): lista = [] i = 1 while i < l: a = fibonacci(i) if a % 2 == 0: lista.append(a) i = i + 1 else: i = i + 1 return lista def fibonacci(n): a = 1 b = 2 c = 0 i = 2 if n > 2: while i < n: ...
_____no_output_____
MIT
2-Even Fibonacci numbers(right).ipynb
vss-py/TDD-ProjectEuler
Now, for the next step, I must to set a interval condition. In the statement, the condition is "the values not exced 4 milions".
def fibonacci_list_interval(m): lista = [] i = 1 while True: a = fibonacci(i) if a > m: print('The last term is: ' + str(a)) break if a % 2 == 0: lista.append(a) i = i + 1 else: i = i + 1 return lista def fibon...
The last term is: 5702887
MIT
2-Even Fibonacci numbers(right).ipynb
vss-py/TDD-ProjectEuler
I don't know how can I make first a test in this case, but how the code was developed, I see no one problem to do this way. Whatever, the new function is just the previous function with same changes. To return the sum of even numbers of the fibonacci list, I must to do only this:
def fibonacci_list_interval(m): lista = [] i = 1 while True: a = fibonacci(i) if a > m: print('The last term is: ' + str(a)) break if a % 2 == 0: lista.append(a) i = i + 1 else: i = i + 1 return sum(lista) def ...
The last term is: 5702887
MIT
2-Even Fibonacci numbers(right).ipynb
vss-py/TDD-ProjectEuler
Chapter 3 - Regression Models Segment 2 - Multiple linear regression
import numpy as np import pandas as pd import matplotlib.pyplot as plt from pylab import rcParams import sklearn from sklearn.linear_model import LinearRegression from sklearn.preprocessing import scale %matplotlib inline rcParams['figure.figsize'] = 5, 4 import seaborn as sb sb.set_style('whitegrid') from collection...
_____no_output_____
MIT
modules/module_12/part3/03_02_begin/03_02_begin.ipynb
ryanjmccall/sb_ml_eng_capstone
(Multiple) linear regression on the enrollment data
address = 'C:/Users/Lillian/Desktop/ExerciseFiles/Data/enrollment_forecast.csv' enroll = pd.read_csv(address) enroll.columns = ['year', 'roll', 'unem', 'hgrad', 'inc'] enroll.head() sb.pairplot(enroll) print(enroll.corr()) enroll_data = enroll[['unem', 'hgrad']].values enroll_target = enroll[['roll']].values enroll_...
_____no_output_____
MIT
modules/module_12/part3/03_02_begin/03_02_begin.ipynb
ryanjmccall/sb_ml_eng_capstone
Checking for missing values
missing_values = X==np.NAN X[missing_values == True] LinReg = LinearRegression(normalize=True) LinReg.fit(X, y) print(LinReg.score(X, y))
0.8488812666133723
MIT
modules/module_12/part3/03_02_begin/03_02_begin.ipynb
ryanjmccall/sb_ml_eng_capstone
Discrete Models Task 1) Data Preparation Installing 'tslearn' - only run if not installed previously (if necessary, for more information see documentation of the package at https://tslearn.readthedocs.io/en/latest/index.html?highlight=install
##!pip install tslearn
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Importing packages
import datetime import pandas as pd from pandas import Series from pandas import DataFrame from pandas import concat import numpy as np import time, datetime import matplotlib.pyplot as plt import sys from tslearn.generators import random_walks from tslearn.preprocessing import TimeSeriesScalerMeanVariance from tslearn...
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Reading the data, converting the time to timestamp and indexing the date to use it as a time series
#Data Paths DATA_PATH = sys.path[0]+"\\data\\" filename_train1="BATADAL_dataset03.csv" filename_train2="BATADAL_dataset04.csv" filename_test="BATADAL_test_dataset.csv" #Reading the data dftrain1 = pd.read_csv(DATA_PATH + filename_train1) dftrain2 = pd.read_csv(DATA_PATH + filename_train2) dftest = pd.read_csv(DATA_PAT...
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Scaling the data to normal distribution with mean zero and standard deviation
#Obs: some values will be converted to float dftrain1_original=dftrain1 dftrain2_original=dftrain2 dftest_original=dftest #Dftrain1 dftrain1.iloc[:,range(0,31)]=preprocessing.scale(dftrain1.iloc[:,range(0,31)]) #Dftrain2 dftrain2.iloc[:,range(0,31)]=preprocessing.scale(dftrain2.iloc[:,range(0,31)]) #Dftest dftest.iloc[...
C:\Users\pvbia\Anaconda3\lib\site-packages\ipykernel_launcher.py:6: DataConversionWarning: Data with input dtype int64, float64 were all converted to float64 by the scale function. C:\Users\pvbia\Anaconda3\lib\site-packages\ipykernel_launcher.py:10: DataConversionWarning: Data with input dtype int64, float64 were al...
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Labeling the attacks on dftrain2 according to the information below For L_T713/09/2016 23 - 16/09/2016 00 |26/09/2016 11 - 27/09/2016 10 For F_PU10 and F_PU1126/09/2016 11 - 27/09/2016 10 For L_T109/10/2016 9 - 11/10/2016 20 |29/10/2016 19 - 02/11/2016 16 |14/12/2016 15 - 19/12/2016 04 For F_PU1 and F_PU229/10/2016 19...
#L_T7 dftrain2_L_T7=dftrain2.loc[:,[' L_T7',' ATT_FLAG']] dftrain2_L_T7.loc[:,' ATT_FLAG']=0 dftrain2_L_T7.loc['2016-09-13 23':'2016-09-16 00',' ATT_FLAG']=1 dftrain2_L_T7.loc['2016-09-26 11':'2016-09-27 10',' ATT_FLAG']=1 #F_PU10 and F_PU11 dftrain2_F_PU10=dftrain2.loc[:,[' F_PU10',' ATT_FLAG']] dftrain2_F_PU10.loc[:,...
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Labling the attacks on dftest according to the batadal dataset
###Labling zero in the test set for all anomalies dftest[" ATT_FLAG"]=0 ###Creating one dataset for each dimension #Attack 1 dftest_L_T3=dftest.loc[:,['L_T3',' ATT_FLAG']] dftest_P_U4=dftest.loc[:,['F_PU4', ' ATT_FLAG']] dftest_P_U5=dftest.loc[:,['F_PU5', ' ATT_FLAG']] #Attack 2 dftest_L_T2=dftest.loc[:,['L_T2', ' ATT_...
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Labling all the attacks in one dataset to compute performance
dftest.loc['2017-01-16 09':'2017-01-19 06',' ATT_FLAG']=1 dftest.loc['2017-01-30 08':'2017-02-02 00',' ATT_FLAG']=1 dftest.loc['2017-02-09 03':'2017-02-10 09',' ATT_FLAG']=1 dftest.loc['2017-02-12 01':'2017-02-13 07',' ATT_FLAG']=1 dftest.loc['2017-02-24 05':'2017-02-28 08',' ATT_FLAG']=1 dftest.loc['2017-03-10 14':'20...
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Adding the anomaly type column to be used in the detection
dftest["anomalytype"]=0
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Attack 1
plt.plot(dftest_L_T3.loc['2017-01-16 09':'2017-01-19 06', 'L_T3'],color="red") plt.plot(dftest_L_T3.loc['2017-01-12 09':'2017-01-16 09', 'L_T3'],color="blue") plt.plot(dftest_L_T3.loc['2017-01-19 06':'2017-01-23 06', 'L_T3'],color="blue") plt.plot(dftest_P_U4.loc['2017-01-16 09':'2017-01-19 06', 'F_PU4'],color="red") ...
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Thus attack 1 generated a contextual anomaly in L_T3
dftest.loc['2017-01-16 09':'2017-01-19 06', 'anomalytype']="contextual"
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Attack 2
plt.plot(dftest_L_T2.loc['2017-01-30 08':'2017-02-02 00', 'L_T2'],color="red") plt.plot(dftest_L_T2.loc['2017-01-26 08':'2017-01-30 08', 'L_T2'],color="blue") plt.plot(dftest_L_T2.loc['2017-02-02 00':'2017-02-06 00', 'L_T2'],color="blue")
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Thus attack 2 generated a collective anomaly in L_T2
dftest.loc['2017-01-30 08':'2017-02-02 00', 'anomalytype']="collective"
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Attack 3
plt.plot(dftest_P_U3.loc['2017-02-09 02':'2017-02-10 09', 'F_PU3'],color="red") plt.plot(dftest_P_U3.loc['2017-02-05 03':'2017-02-09 02', 'F_PU3'],color="blue") plt.plot(dftest_P_U3.loc['2017-02-10 09':'2017-02-14 09', 'F_PU3'],color="blue")
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Thus attack 3 generated a collective anomaly in F_PU3
dftest.loc['2017-02-09 02':'2017-02-10 09', 'anomalytype']="collective"
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Attack 4
plt.plot(dftest_P_U3.loc['2017-02-12 00':'2017-02-13 07', 'F_PU3'],color="red") plt.plot(dftest_P_U3.loc['2017-02-08 01':'2017-02-12 00', 'F_PU3'],color="blue") plt.plot(dftest_P_U3.loc['2017-02-13 07':'2017-02-17 07', 'F_PU3'],color="blue")
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Thus attack 4 generated a collective anomaly in F_PU3
dftest.loc['2017-02-12 00':'2017-02-13 07', 'anomalytype']="collective"
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Attack 5
plt.plot(dftest_L_T2.loc['2017-02-24 05':'2017-02-28 08', 'L_T2'],color="red") plt.plot(dftest_L_T2.loc['2017-02-20 05':'2017-02-24 05', 'L_T2'],color="blue") plt.plot(dftest_L_T2.loc['2017-02-28 08':'2017-03-03 08', 'L_T2'],color="blue") plt.plot(dftest_V2.loc['2017-02-24 05':'2017-02-28 08', 'F_V2'],color="red") pl...
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Thus attack 5 generated a collective anomaly in L_T2, F_V2, P_J14 and P_J422
dftest.loc['2017-02-24 05':'2017-02-28 08', 'anomalytype']="collective"
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Attack 6
plt.plot(dftest_L_T7.loc['2017-03-10 14':'2017-03-13 21', 'L_T7'],color="red") plt.plot(dftest_L_T7.loc['2017-03-06 14':'2017-03-10 14', 'L_T7'],color="blue") plt.plot(dftest_L_T7.loc['2017-03-13 21':'2017-03-17 21', 'L_T7'],color="blue") plt.plot(dftest_P_U10.loc['2017-03-10 14':'2017-03-13 21', 'F_PU10'],color="red"...
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Attack 6 generated collective anomalies
dftest.loc['2017-03-10 14':'2017-03-13 21', 'anomalytype']="collective"
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Attack 7
plt.plot(dftest_L_T4.loc['2017-03-25 20':'2017-03-27 01', 'L_T4'],color="red") plt.plot(dftest_L_T4.loc['2017-03-21 20':'2017-03-25 20', 'L_T4'],color="blue") plt.plot(dftest_L_T4.loc['2017-03-27 01':'2017-03-31 01', 'L_T4'],color="blue") plt.plot(dftest_L_T6.loc['2017-03-25 20':'2017-03-27 01', 'L_T6'],color="red") ...
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Thus attack 6 generated a collective anomaly in L_T4 and L_T6
dftest.loc['2017-03-25 20':'2017-03-27 01', 'anomalytype']="collective"
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
2) Discretization Creating a dataset to test the sax
dataset_scale=dftrain1['L_T7']
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Creating the Symbolic Aggregate Approximation (SAX) function and testing in the above mentioned dataset
### Creting a function that creates the sax def tosax(n_paa_segments,n_sax_symbols,dataset_scale): #Creating the SAX using 'n_paa_segments' as number of segments and 'n_sax_symbols' as number of symbols sax = SymbolicAggregateApproximation(n_segments=n_paa_segments, alphabet_size_avg=n_sax_symbols) # Comput...
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Plotting the scaled time series, PAA and SAX representations for the sake of validation of the code
#Plotting - only 100 first values to test the code isworking as expected # Plotting the scaled data set plt.subplot(2, 2, 2) # First, raw time series plt.plot(dataset_scale.ravel()[0:300], "b-") plt.title("Original Time Series") #Plotting the SAX plt.subplot(2, 2, 1) # Then SAX plt.plot(sax_dataset_inv[0].ravel()[0...
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
3) N-Grams Representations Creating a function that returns its n-gram representation.
#'a' will be used just to test the functions below. a=sax_dataset_inv[0].ravel()
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
We need function(s) to convert numeric values generated from the sax to a string to perform N-grams as a it was a NLP problem. For this purpose two functions were created: the first just put all the values of a list in a string and the second convert numerical values of the sax to letters
# First function convert a list of numeric values to a string def convert(list): # Converting integer list to string list s = [str(i) for i in list] # Join list items using join() res = ("".join(s)) return(res) # Second function to convert a sax numeric representation into strings def saxtostr...
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Now to generate Ngrams we need a function that gather from the data the possible grams and their possible next values.
#Function that creates the Ngrams representation by creating a dictionary that has as keys the representations and as values #the number elements generated after the n-gram def tongrams(tokens,n): #Creates a dictionary with the Ngrams representation and the next element (that is to be predicted by the ngrams) D...
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Now we create a Markov Chain that calculates the prior and the posterior given the dictionary generated by the previous function.
#Function that calcualte the likelihood given a ngram def tomarkovchain(ngram): #Create dictionary to be filled with the probability of conditional events (likelihood) l_probs=dict() #Create the dictionaries to be filled with the counting for each gram c1=dict() #Loop through keys to calculate the l...
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Lastly, we create a function that loop through the dataframe and return the index that are below a certain threshold
def prediction(markovchain_model,n_sax_symbols,string_sax, threshold): #Calculating number of ngrams used n=len(random.choice(list(markovchain_model[0].keys()))) #Calculationg the number of times a observation should be reproduced so the data set has the same size as the original #Creating the list to ...
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
And a function that merge the predictions into the original dataframe considering the possible rounding differences that may lead to difference in length when creating SAX
def smartmerge(df,predictions): df=pd.Series.to_frame(df) if len(predictions)==len(df): df.insert(1, "predictions", predictions) df.reset_index(inplace=True) if len(predictions)>len(df): predictions=predictions[:-(len(predictions)-len(df))] df.insert(1, "predictions...
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
4) Predictions Attacks Using N-grams _In this subchapter we are going to try to predict the attacks on the training dataset 2_ Creating functions that aid the prediction Creating a function that make predicitons and calculate performance indicators.
def model_to_predict(trainset,testset,results,n_sax_symbols,n_paa_segments,n_grams,threshold): ###Generating the model for the train set### #Sax sax_train=tosax(n_paa_segments,n_sax_symbols,trainset).ravel() #Sax to string sax_string_train=saxtostring(sax_train) #Ngrams ngrams=tongrams(s...
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Creating a function that plot the confusion matrix in a nice format!
def ploting_cm(cm): #Plot ax= plt.subplot() sns.heatmap(cm, annot=True,fmt='g', ax = ax); #annot=True to annotate cells # labels, title and ticks ax.set_xlabel('Predicted labels');ax.set_ylabel('True labels'); ax.set_title('Confusion Matrix'); ax.xaxis.set_ticklabels(['normal', 'anom...
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics
Creating a function that loop over the parameters and select the best set of parameters based the F-score
def parameterselection(trainset,testset,results): #Genereting one first model to have a starting value for the loop with selected parameters parameters={"sax":5,"paa":1,"ngram":3,"threshold":0.10} best_MODEL=model_to_predict(trainset,testset,results,5,int(len(trainset)/1),3,0.1) #Looping for sax in ...
_____no_output_____
MIT
batadal/Task3_Discrete.ipynb
imchengh/Cyber-Data-Analytics