markdown stringlengths 0 37k | code stringlengths 1 33.3k | path stringlengths 8 215 | repo_name stringlengths 6 77 | license stringclasses 15
values |
|---|---|---|---|---|
调整gamma参数 | svm = SVC(kernel='rbf', random_state=0, gamma=100.0, C=1.0)
svm.fit(X_train_std, y_train)
plot_decision_regions(X_combined_std, y_combined, classifier=svm, test_idx=range(105,150))
plt.xlabel('petal length [standardized]')
plt.ylabel('petal width [standardized]')
plt.legend(loc='upper left')
plt.show() | jupyter/machine_learning_1.ipynb | lichao890427/lichao890427.github.io | mit |
1.5 使用sklearn决策树实现分类器
  如果我们关心的是模型的可解释性，决策树是有用的模型，通过不断问问题走向不同的分支，最终得到类型 | from sklearn.tree import DecisionTreeClassifier
tree = DecisionTreeClassifier(criterion='entropy', max_depth=3, random_state=0)
tree.fit(X_train, y_train)
X_combined = np.vstack((X_train, X_test))
y_combined = np.hstack((y_train, y_test))
plot_decision_regions(X_combined, y_combined, classifier=tree, test_idx=range(105... | jupyter/machine_learning_1.ipynb | lichao890427/lichao890427.github.io | mit |
1.6 使用随机森林实现分类器
  随机森林由于在实现分类时效果良好，可伸缩，易用的特性比较受欢迎。随机森林可以看作决策树的组合，弱弱联合组成更健壮的模型，降低决策树的过拟合。随机森林算法可以总结为4步：
* 从样本集中通过重采样的方式产生n个样本
* 假设样本特征数目为a，对n个样本选择a中的k个特征，用建立决策树的方式获得最佳分割点
* 重复m次，产生m棵决策树
* 多数投票机制来进行预测
  随机森林的优势在于无需考虑如何选择超参数，模型足够健壮无需剪枝，只需要考虑决策树的个数k，k越大表现越良好但是计算成本就越大。 | from sklearn.ensemble import RandomForestClassifier
forest = RandomForestClassifier(criterion='entropy', n_estimators=10, random_state=1, n_jobs=2)
forest.fit(X_train, y_train)
plot_decision_regions(X_combined, y_combined, classifier=forest, test_idx=range(105,150))
plt.xlabel('petal length')
plt.ylabel('petal width')
... | jupyter/machine_learning_1.ipynb | lichao890427/lichao890427.github.io | mit |
1.7 K近邻算法实现分类器
  这是一个监督机器学习算法(KNN)，KNN是典型的懒惰学习算法，他会记忆训练数据集而不是从数据集得到判断函数
参数模型和非参数模型
  机器学习算法可分为参数模型和非参数模型。参数模型用于从训练集估计参数，产生的函数无需用到以前的数据就可以为新数据分类，典型的例子是感知器、逻辑回归、线性SVM；非参数模型无法从固定参数集合获取特征，且参数数量随着训练集增加而增加，典型的例子是决策树、随机森林和核函数SVM、K近邻 | from sklearn.neighbors import KNeighborsClassifier
knn = KNeighborsClassifier(n_neighbors=5, p=2, metric='minkowski')
knn.fit(X_train_std, y_train)
plot_decision_regions(X_combined_std, y_combined, classifier=knn, test_idx=range(105,150))
plt.xlabel('petal length [standardized]')
plt.ylabel('petal width [standardized]'... | jupyter/machine_learning_1.ipynb | lichao890427/lichao890427.github.io | mit |
减小过拟合
收集更多数据
通过正规化引入复杂度惩罚(L1正规化)
使用更少参数构建简单一些的模型
降维(序列化特征选择)
使用L1正规化 | import pandas as pd
from sklearn.preprocessing import StandardScaler
from sklearn.cross_validation import train_test_split
df_wine = pd.read_csv('https://archive.ics.uci.edu/ml/machine-learning-databases/wine/wine.data', header=None)
df_wine.columns = ['Class label', 'Alcohol', 'Malic acid', 'Ash', 'Alcalinity of ash',... | jupyter/machine_learning_1.ipynb | lichao890427/lichao890427.github.io | mit |
使用SBS序列化特征选择 | from sklearn.base import clone
from itertools import combinations
import numpy as np
from sklearn.cross_validation import train_test_split
from sklearn.metrics import accuracy_score
class SBS():
def __init__(self, estimator, k_features, scoring=accuracy_score, test_size=0.25, random_state=1):
self.scoring =... | jupyter/machine_learning_1.ipynb | lichao890427/lichao890427.github.io | mit |
我们挑选5个特征检查是否带来改善,从结果可以看出，使用更少的属性，测试集的准确率提高了2% | k5 = list(sbs.subsets_[8])
print(df_wine.columns[1:][k5])
knn.fit(X_train_std, y_train)
print('Training accuracy:', knn.score(X_train_std, y_train))
print('Test accuracy:', knn.score(X_test_std, y_test))
knn.fit(X_train_std[:, k5], y_train)
print('Training accuracy(select 5):', knn.score(X_train_std[:, k5], y_train))
p... | jupyter/machine_learning_1.ipynb | lichao890427/lichao890427.github.io | mit |
使用随机森林评估特征重要程度
  前面我们用L1标准化去除不相关特征，用SBS算法选择特征。另一种选择相关特性的方式是随机森林。 | from sklearn.ensemble import RandomForestClassifier
feat_labels = df_wine.columns[1:]
forest = RandomForestClassifier(n_estimators=10000, random_state=0, n_jobs=-1)
forest.fit(X_train, y_train)
importances = forest.feature_importances_
indices = np.argsort(importances)[::-1]
for f in range(X_train.shape[1]):
print(... | jupyter/machine_learning_1.ipynb | lichao890427/lichao890427.github.io | mit |
2.1 通过降维压缩数据
主特征分析(PCA)，压缩非监督学习数据
线性判别分析(PDA)，降维监督学习数据
核心主特征分析(KPCA)，降维非线性数据
Sklearn PCA | from matplotlib.colors import ListedColormap
from sklearn.linear_model import LogisticRegression
from sklearn.decomposition import PCA
import matplotlib.pyplot as plt
import pandas as pd
from sklearn.cross_validation import train_test_split
from sklearn.preprocessing import StandardScaler
def plot_decision_regions(X, ... | jupyter/machine_learning_1.ipynb | lichao890427/lichao890427.github.io | mit |
Build the network
For the neural network, you'll build each layer into a function. Most of the code you've seen has been outside of functions. To test your code more thoroughly, we require that you put each layer in a function. This allows us to give you better feedback and test for simple mistakes using our unittest... | import tensorflow as tf
def neural_net_image_input(image_shape):
"""
Return a Tensor for a bach of image input
: image_shape: Shape of the images
: return: Tensor for image input.
"""
# TODO: Implement Function
return None
def neural_net_label_input(n_classes):
"""
Return a Tensor... | project2/files/dlnd_image_classification_instruction.ipynb | myfunprograms/deep_learning | gpl-3.0 |
Import packages | # Import
from __future__ import absolute_import, division, print_function
import calendar
import hashlib
import json
import math
import os
import random
import time
import uuid
from datetime import datetime
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import seaborn as sns
from google.cloud ... | notebooks/community/analytics-componetized-patterns/retail/ltv/bqml/notebooks/bqml_automl_ltv_activate_lookalike.ipynb | GoogleCloudPlatform/bigquery-notebooks | apache-2.0 |
Authenticate your GCP account
If you are using AI Platform Notebooks, you are already authenticated so there is no need to run this step. | import sys
if "google.colab" in sys.modules:
from google.colab import auth as google_auth
google_auth.authenticate_user() | notebooks/community/analytics-componetized-patterns/retail/ltv/bqml/notebooks/bqml_automl_ltv_activate_lookalike.ipynb | GoogleCloudPlatform/bigquery-notebooks | apache-2.0 |
Analyze dataset
Some charts might use a log scale.
Quantity
This sections shows how to use the BigQuery ML BUCKETIZE preprocessing function to create buckets of data for quantity and display a log scaled distribution of the qty field. | %%bigquery df_histo_qty --project $PROJECT_ID
WITH
min_max AS (
SELECT
MIN(qty) min_qty,
MAX(qty) max_qty,
CEIL((MAX(qty) - MIN(qty)) / 100) step
FROM
`ltv_ecommerce.10_orders`
)
SELECT
COUNT(1) c,
bucket_same_size AS bucket
FROM (
SELECT
-- Creates (1000-100)/100 + 1 buckets of data.
... | notebooks/community/analytics-componetized-patterns/retail/ltv/bqml/notebooks/bqml_automl_ltv_activate_lookalike.ipynb | GoogleCloudPlatform/bigquery-notebooks | apache-2.0 |
Unit price | %%bigquery df_histo_unit_price --project $PROJECT_ID
WITH
min_max AS (
SELECT
MIN(unit_price) min_unit_price,
MAX(unit_price) max_unit_price,
CEIL((MAX(unit_price) - MIN(unit_price)) / 10) step
FROM
`ltv_ecommerce.10_orders`
)
SELECT
COUNT(1) c,
bucket_same_size AS bucket
FROM (
SELECT
... | notebooks/community/analytics-componetized-patterns/retail/ltv/bqml/notebooks/bqml_automl_ltv_activate_lookalike.ipynb | GoogleCloudPlatform/bigquery-notebooks | apache-2.0 |
Set parameters for LTV
Some parameters useful to run some of the queries in this tutorial:
WINDOW_STEP: How many days between threshold dates.
WINDOW_STEP_INITIAL: How many days between the first order and the first threshold date. A threshold date is when BigQuery computes inputs and targets.
WINDOW_LENGTH: How many ... | LTV_PARAMS = {
"WINDOW_LENGTH": 0,
"WINDOW_STEP": 30,
"WINDOW_STEP_INITIAL": 90,
"LENGTH_FUTURE": 30,
"MAX_STDV_MONETARY": 500,
"MAX_STDV_QTY": 100,
"TOP_LTV_RATIO": 0.2,
}
LTV_PARAMS | notebooks/community/analytics-componetized-patterns/retail/ltv/bqml/notebooks/bqml_automl_ltv_activate_lookalike.ipynb | GoogleCloudPlatform/bigquery-notebooks | apache-2.0 |
Check distributions
This tutorial does minimum data cleansing and focuses mostly on transforming a list of transactions into workable inputs for the model.
This section checks that data is generally usable.
Per date | %%bigquery df_dist_dates --project $PROJECT_ID
SELECT count(1) c, SUBSTR(CAST(order_day AS STRING), 0, 7) as yyyy_mm
FROM `ltv_ecommerce.20_aggred`
WHERE qty_articles > 0
GROUP BY yyyy_mm
ORDER BY yyyy_mm
plt.figure(figsize=(12, 5))
sns.barplot(x="yyyy_mm", y="c", data=df_dist_dates) | notebooks/community/analytics-componetized-patterns/retail/ltv/bqml/notebooks/bqml_automl_ltv_activate_lookalike.ipynb | GoogleCloudPlatform/bigquery-notebooks | apache-2.0 |
Orders are quite well distributed across the year despite a lower number in the early days of the dataset. You can keep this in mind when choosing a value for WINDOW_STEP_INITIAL.
Per customer | %%bigquery df_dist_customers --params $LTV_PARAMS --project $PROJECT_ID
SELECT customer_id, count(1) c
FROM `ltv_ecommerce.20_aggred`
GROUP BY customer_id
plt.figure(figsize=(12, 4))
sns.distplot(df_dist_customers["c"], hist_kws=dict(ec="k"), kde=False) | notebooks/community/analytics-componetized-patterns/retail/ltv/bqml/notebooks/bqml_automl_ltv_activate_lookalike.ipynb | GoogleCloudPlatform/bigquery-notebooks | apache-2.0 |
The number of transactions per customer is distributed across a few discrete values with no clear outliers.
Per quantity
This section looks at the general distribution of the number of articles per orders and check if there are some outliers. | %%bigquery df_dist_qty --params $LTV_PARAMS --project $PROJECT_ID
SELECT qty_articles, count(1) c
FROM `ltv_ecommerce.20_aggred`
GROUP BY qty_articles
plt.figure(figsize=(12, 4))
sns.distplot(df_dist_qty["qty_articles"], hist_kws=dict(ec="k"), kde=False) | notebooks/community/analytics-componetized-patterns/retail/ltv/bqml/notebooks/bqml_automl_ltv_activate_lookalike.ipynb | GoogleCloudPlatform/bigquery-notebooks | apache-2.0 |
Dataset | %%bigquery --project $PROJECT_ID
-- Shows all data for a specific customer and some other random records.
SELECT * FROM `ltv_ecommerce.30_featured` WHERE customer_id = "10"
UNION ALL
(SELECT * FROM `ltv_ecommerce.30_featured` LIMIT 5)
ORDER BY customer_id, frequency, T
%%bigquery df_featured --project $PROJECT_ID
lt... | notebooks/community/analytics-componetized-patterns/retail/ltv/bqml/notebooks/bqml_automl_ltv_activate_lookalike.ipynb | GoogleCloudPlatform/bigquery-notebooks | apache-2.0 |
Seems like for most values, there is a long tail of records. This is something that might required additional feature preparation even if AutoML already provides some automatic engineering. You can investigate this if you want to improve the base model.
Train the model
This tutorial uses an AutoML regressor to predict ... | # You can run this query using the magic cell but the cell would run for hours.
# Although stopping the cell would not stop the query, using the Python client
# also enables you to add a custom parameter for the model name.
suffix_now = datetime.now().strftime("%Y%m%d_%H%M%S")
train_model_jobid = f"train_model_{suffix_... | notebooks/community/analytics-componetized-patterns/retail/ltv/bqml/notebooks/bqml_automl_ltv_activate_lookalike.ipynb | GoogleCloudPlatform/bigquery-notebooks | apache-2.0 |
This is an example of a model evaluation
Predict LTV
Predicts LTV for all customers. It uses the overall monetary value for each customer to predict a future one. | %%bigquery --params $LTV_PARAMS --project $PROJECT_ID
-- TODO(developer):
-- 1. Update the model name to the one you want to use.
-- 2. Update the table where to output predictions.
-- How many days back for inputs transactions. 0 means from the start.
DECLARE WINDOW_LENGTH INT64 DEFAULT @WINDOW_LENGTH;
-- Date at w... | notebooks/community/analytics-componetized-patterns/retail/ltv/bqml/notebooks/bqml_automl_ltv_activate_lookalike.ipynb | GoogleCloudPlatform/bigquery-notebooks | apache-2.0 |
The monetary distribution analysis shows small monetary amounts for the next month compare to the overall historical value. The difference is about 3 to 4 orders of magnitude.
One reason is that the model is trained to predict the value for the next month (LENGTH_FUTURE = 30).
You can play around with that value to tr... | %%bigquery df_top_ltv --params $LTV_PARAMS --project $PROJECT_ID
DECLARE TOP_LTV_RATIO FLOAT64 DEFAULT @TOP_LTV_RATIO;
SELECT
p.customer_id,
monetary_future,
c.email AS email
FROM (
SELECT
customer_id,
monetary_future,
PERCENT_RANK() OVER (ORDER BY monetary_future DESC) AS percent_rank_monetary
... | notebooks/community/analytics-componetized-patterns/retail/ltv/bqml/notebooks/bqml_automl_ltv_activate_lookalike.ipynb | GoogleCloudPlatform/bigquery-notebooks | apache-2.0 |
Setup Adwords client
Creates the configuration YAML file for the Google Ads client. You need to:
1. Create Client ID and Secret using the Cloud Console
2. Follow these steps | # Sets your variables.
if "google.colab" in sys.modules:
from google.colab import files
ADWORDS_FILE = "/tmp/adwords.yaml"
DEVELOPER_TOKEN = "[YOUR_DEVELOPER_TOKEN]"
OAUTH_2_CLIENT_ID = "[YOUR_OAUTH_2_CLIENT_ID]"
CLIENT_SECRET = "[YOUR_CLIENT_SECRET]"
REFRESH_TOKEN = "[YOUR_REFRESH_TOKEN]"
# Creates a local YAML... | notebooks/community/analytics-componetized-patterns/retail/ltv/bqml/notebooks/bqml_automl_ltv_activate_lookalike.ipynb | GoogleCloudPlatform/bigquery-notebooks | apache-2.0 |
Create an AdWords user list
Using emails of the top LTV customers, you create an AdWords list. If more than 5000 of the users are matched with AdWords email, a similar audience list will be created.
Note that this guide uses fake emails so running these steps is not going to work but you can leverage this code with em... | ltv_emails = list(set(df_top_ltv["email"]))
# https://developers.google.com/adwords/api/docs/samples/python/remarketing#create-and-populate-a-user-list
# https://github.com/googleads/googleads-python-lib/blob/7c41584c65759b6860572a13bde65d7395c5b2d8/examples/adwords/v201809/remarketing/add_crm_based_user_list.py
# ""... | notebooks/community/analytics-componetized-patterns/retail/ltv/bqml/notebooks/bqml_automl_ltv_activate_lookalike.ipynb | GoogleCloudPlatform/bigquery-notebooks | apache-2.0 |
SWAP | swap() | examples/quantum-gates.ipynb | ajgpitch/qutip-notebooks | lgpl-3.0 |
ISWAP | iswap() | examples/quantum-gates.ipynb | ajgpitch/qutip-notebooks | lgpl-3.0 |
From QuTiP 4.4, we can also add gate at arbitrary position in a circuit. | qc1.add_gate("CSIGN", index=1)
qc1.png | examples/quantum-gates.ipynb | ajgpitch/qutip-notebooks | lgpl-3.0 |
Adding gate in the middle of a circuit
From QuTiP 4.4 one can add a gate at an arbitrary position of a circuit. All one needs to do is to specify the parameter index. With this, we can also add the same gate at multiple positions at the same time. | qc = QubitCircuit(1)
qc.add_gate("RX", targets=1)
qc.add_gate("RX", targets=1)
qc.add_gate("RY", targets=1, index=[1,0])
qc.gates | examples/quantum-gates.ipynb | ajgpitch/qutip-notebooks | lgpl-3.0 |
If
If statements can be use to execute some lines or block of code if a particular condition is satisfied. E.g. Let's print something based on the entries in the list. | for instructor in instructors:
if not "Clown" in instructor:
print(instructor)
else:
pass
for instructor in instructors:
if "Clown" in instructor:
pass
else:
print(instructor)
for instructor in instructors:
if not "Clown" in instructor:
print(instructor)
... | week_1/procedural_python/flow_of_control.ipynb | UWSEDS/LectureNotes | bsd-2-clause |
You can combine loops and conditionals: | for instructor in instructors:
if instructor.endswith('Clown'):
print(instructor + " doesn't sound like a real instructor name!")
else:
print(instructor + " is so smart... all those gooey brains!")
# Loops can be nested
for i in range(1, 4):
for j in range(1, 4):
print('%d * %d = %d... | week_1/procedural_python/flow_of_control.ipynb | UWSEDS/LectureNotes | bsd-2-clause |
Programming Example
Write a script that finds the first N prime numbers. | 4 % 2
N = 100
for candidate in range(2, N):
# n is candidate prime. Check if n is prime
is_prime = True
for m in range(2, candidate):
if (candidate % m) == 0:
is_prime = False
break
if is_prime:
print("%d is prime!" % candidate)
| week_1/procedural_python/flow_of_control.ipynb | UWSEDS/LectureNotes | bsd-2-clause |
With the model loaded, you can process text like this: | doc = nlp("Tea is healthy and calming, don't you think?") | notebooks/nlp/raw/tut1.ipynb | Kaggle/learntools | apache-2.0 |
There's a lot you can do with the doc object you just created.
Tokenizing
This returns a document object that contains tokens. A token is a unit of text in the document, such as individual words and punctuation. SpaCy splits contractions like "don't" into two tokens, "do" and "n't". You can see the tokens by iterating ... | for token in doc:
print(token) | notebooks/nlp/raw/tut1.ipynb | Kaggle/learntools | apache-2.0 |
Iterating through a document gives you token objects. Each of these tokens comes with additional information. In most cases, the important ones are token.lemma_ and token.is_stop.
Text preprocessing
There are a few types of preprocessing to improve how we model with words. The first is "lemmatizing."
The "lemma" of a w... | print(f"Token \t\tLemma \t\tStopword".format('Token', 'Lemma', 'Stopword'))
print("-"*40)
for token in doc:
print(f"{str(token)}\t\t{token.lemma_}\t\t{token.is_stop}") | notebooks/nlp/raw/tut1.ipynb | Kaggle/learntools | apache-2.0 |
Why are lemmas and identifying stopwords important? Language data has a lot of noise mixed in with informative content. In the sentence above, the important words are tea, healthy and calming. Removing stop words might help the predictive model focus on relevant words. Lemmatizing similarly helps by combining multiple ... | from spacy.matcher import PhraseMatcher
matcher = PhraseMatcher(nlp.vocab, attr='LOWER') | notebooks/nlp/raw/tut1.ipynb | Kaggle/learntools | apache-2.0 |
The matcher is created using the vocabulary of your model. Here we're using the small English model you loaded earlier. Setting attr='LOWER' will match the phrases on lowercased text. This provides case insensitive matching.
Next you create a list of terms to match in the text. The phrase matcher needs the patterns as ... | terms = ['Galaxy Note', 'iPhone 11', 'iPhone XS', 'Google Pixel']
patterns = [nlp(text) for text in terms]
matcher.add("TerminologyList", patterns) | notebooks/nlp/raw/tut1.ipynb | Kaggle/learntools | apache-2.0 |
Then you create a document from the text to search and use the phrase matcher to find where the terms occur in the text. | # Borrowed from https://daringfireball.net/linked/2019/09/21/patel-11-pro
text_doc = nlp("Glowing review overall, and some really interesting side-by-side "
"photography tests pitting the iPhone 11 Pro against the "
"Galaxy Note 10 Plus and last year’s iPhone XS and Google Pixel 3.")
matc... | notebooks/nlp/raw/tut1.ipynb | Kaggle/learntools | apache-2.0 |
The matches here are a tuple of the match id and the positions of the start and end of the phrase. | match_id, start, end = matches[0]
print(nlp.vocab.strings[match_id], text_doc[start:end]) | notebooks/nlp/raw/tut1.ipynb | Kaggle/learntools | apache-2.0 |
We initialize the simulation and generate the grid
in the complex plane. | size = 200
iterations = 100 | 4_Cython.ipynb | thewtex/ieee-nss-mic-scipy-2014 | apache-2.0 |
Pure Python | def mandelbrot_python(m, size, iterations):
for i in range(size):
for j in range(size):
c = -2 + 3./size*j + 1j*(1.5-3./size*i)
z = 0
for n in range(iterations):
if np.abs(z) <= 10:
z = z*z + c
m[i, j] = n
... | 4_Cython.ipynb | thewtex/ieee-nss-mic-scipy-2014 | apache-2.0 |
Cython versions
We first import Cython. | %load_ext cythonmagic | 4_Cython.ipynb | thewtex/ieee-nss-mic-scipy-2014 | apache-2.0 |
Take 1
First, we just add the %%cython magic. | %%cython -a
import numpy as np
def mandelbrot_cython(m, size, iterations):
for i in range(size):
for j in range(size):
c = -2 + 3./size*j + 1j*(1.5-3./size*i)
z = 0
for n in range(iterations):
if np.abs(z) <= 10:
z = z*z + c
... | 4_Cython.ipynb | thewtex/ieee-nss-mic-scipy-2014 | apache-2.0 |
Virtually no speedup.
Take 2
Now, we add type information, using memory views for NumPy arrays. | %%cython -a
import numpy as np
def mandelbrot_cython(int[:,::1] m,
int size,
int iterations):
cdef int i, j, n
cdef complex z, c
for i in range(size):
for j in range(size):
c = -2 + 3./size*j + 1j*(1.5-3./size*i)
z = 0
... | 4_Cython.ipynb | thewtex/ieee-nss-mic-scipy-2014 | apache-2.0 |
Vertex client library: Custom training image classification model with custom container for online prediction
<table align="left">
<td>
<a href="https://colab.research.google.com/github/GoogleCloudPlatform/vertex-ai-samples/blob/master/notebooks/community/gapic/custom/showcase_custom_image_classification_online_c... | import os
import sys
# Google Cloud Notebook
if os.path.exists("/opt/deeplearning/metadata/env_version"):
USER_FLAG = "--user"
else:
USER_FLAG = ""
! pip3 install -U google-cloud-aiplatform $USER_FLAG | notebooks/community/gapic/custom/showcase_custom_image_classification_online_container.ipynb | GoogleCloudPlatform/vertex-ai-samples | apache-2.0 |
Train a model
There are two ways you can train a custom model using a container image:
Use a Google Cloud prebuilt container. If you use a prebuilt container, you will additionally specify a Python package to install into the container image. This Python package contains your code for training a custom model.
Use y... | # Make folder for Python training script
! rm -rf custom
! mkdir custom
# Add package information
! touch custom/README.md
setup_cfg = "[egg_info]\n\ntag_build =\n\ntag_date = 0"
! echo "$setup_cfg" > custom/setup.cfg
setup_py = "import setuptools\n\nsetuptools.setup(\n\n install_requires=[\n\n 'tensorflow... | notebooks/community/gapic/custom/showcase_custom_image_classification_online_container.ipynb | GoogleCloudPlatform/vertex-ai-samples | apache-2.0 |
Select seeds for search networks
I select small (1000-1500) sized bot network and pick 4 random members from it | seeds = ['volya_belousova', 'egor4rgurev', 'kirillfrolovdw', 'ilyazhuchhj']
auth = tweepy.OAuthHandler(OAUTH_KEY, OAUTH_SECRET)
auth.set_access_token(ACCESS_TOKEN, ACCESS_TOKEN_SECRET)
api = tweepy.API(auth, wait_on_rate_limit=True, wait_on_rate_limit_notify=True)
graph = Graph(user=NEO4J_USER, password=NEO4J_SECRET)
... | Twitter bots/Botnet search.ipynb | UserAd/data_science | mit |
Now search for friends of seed users | friend_ids = {}
for account in seeds:
friend_ids[account] = get_friends(account)
commons = {}
for first in seeds:
for second in seeds:
if first != second:
commons[(first, second)] = list(set(friend_ids[first]) & set(friend_ids[second]))
all_users = friend_ids[seeds[0]]
for name in see... | Twitter bots/Botnet search.ipynb | UserAd/data_science | mit |
Show common users in total and per seed user | display("Common users: {0}".format(len(all_users)))
html = ["<table width=100%>"]
html.append('<tr><td></td>')
for name in seeds:
html.append('<td>{0}</td>'.format(name))
html.append('</tr>')
for first in seeds:
html.append('<tr><td>{0}</td>'.format(first))
for second in seeds:
if first != second:... | Twitter bots/Botnet search.ipynb | UserAd/data_science | mit |
Now search and populate neo4j database | graph.run("CREATE CONSTRAINT ON (u:UserRes) ASSERT u.id IS UNIQUE")
processed_users = []
for user_id in all_users:
if user_id not in processed_users:
user = Node("UserRes", id=user_id)
graph.merge(user)
try:
for friend_id in get_follwers_by_id(user_id):
if friend... | Twitter bots/Botnet search.ipynb | UserAd/data_science | mit |
Get all users from neo4j and build graph | query = """
MATCH (user1:UserRes)-[:FRIEND_OF]->(user2:UserRes),
(user2:UserRes)-[:FRIEND_OF]->(user1)
RETURN user1.id, user2.id
"""
data = graph.run(query)
ig = IGraph.TupleList(data, weights=False)
ig.es["width"] = 1
ig.simplify(combine_edges={ "width": "sum" }) | Twitter bots/Botnet search.ipynb | UserAd/data_science | mit |
Let's cluster graph and search for communities | clusters = IGraph.community_fastgreedy(ig)
clusters = clusters.as_clustering()
print("Found %d clusters" % len(clusters)) | Twitter bots/Botnet search.ipynb | UserAd/data_science | mit |
Let's make clusters dataframe | nodes = [{"id": node.index, "name": node["name"]} for node in ig.vs]
for node in nodes:
node["cluster"] = clusters.membership[node["id"]]
nodes_df = pd.DataFrame(nodes)
edges = [{"source": x[0], "target": x[1]} for x in ig.get_edgelist()]
edges_df = pd.DataFrame(edges)
edges_counts = edges_df.groupby('source'... | Twitter bots/Botnet search.ipynb | UserAd/data_science | mit |
Let's look to all clusters closely | nodes_df.groupby('cluster').count() | Twitter bots/Botnet search.ipynb | UserAd/data_science | mit |
We have only two clusters with significant user count.
Let's check first | first_cluster = nodes_df[nodes_df["cluster"] == 0][["id", "name"]] | Twitter bots/Botnet search.ipynb | UserAd/data_science | mit |
Join edges to users | first_cluster_counts = first_cluster.set_index('id').join(edges_counts.set_index('source')).reset_index()
first_cluster_counts["count"].hist() | Twitter bots/Botnet search.ipynb | UserAd/data_science | mit |
Let's look to all groups | for group in range(20):
start = group * 100
stop = (group + 1) * 100
users_slice = first_cluster_counts[(first_cluster_counts["count"] > start) & (first_cluster_counts["count"] < stop)]
print("Users from %d to %d has %d" %(start, stop, users_slice.count()[0]))
display(users_slice[:10]) | Twitter bots/Botnet search.ipynb | UserAd/data_science | mit |
Looks like most bot accounts has followers/follows count from 1200 to 1900
Let's filter it | filtered_bots = first_cluster_counts[(first_cluster_counts["count"] > 1200) & (first_cluster_counts["count"] < 1900)]
print("We found %s bots in first approximation" % filtered_bots.count()[0]) | Twitter bots/Botnet search.ipynb | UserAd/data_science | mit |
Now collect all information from these accounts and search for corellations | first_cluster_bots = []
for group in chunks(filtered_bots["name"].values, 100):
for user in api.lookup_users(user_ids=list(group)):
first_cluster_bots.append(user)
locations = [user.location for user in first_cluster_bots]
first_cluster_bots[0].favourites_count
possible_bot_users = pd.DataFrame([{'name':... | Twitter bots/Botnet search.ipynb | UserAd/data_science | mit |
Ok, we have two significant values. Moscow and New York. Let's split dataset | moscow_users = possible_bot_users[possible_bot_users["location"] == u'Москва']
moscow_users.hist()
moscow_users[:10] | Twitter bots/Botnet search.ipynb | UserAd/data_science | mit |
Now check NY users | ny_users = possible_bot_users[possible_bot_users["location"] == u'New York, USA']
ny_users.hist()
ny_users[:10] | Twitter bots/Botnet search.ipynb | UserAd/data_science | mit |
Conclusion
We have one twitter bot network on two languages: Russian and English.
All bots have deep linking and posts random sentences every hour. | print("Moscow bots: %d, NY bots: %d, Total: %d" % (moscow_users.count()[0], ny_users.count()[0], moscow_users.count()[0] + ny_users.count()[0])) | Twitter bots/Botnet search.ipynb | UserAd/data_science | mit |
Now export moscow and ny users to csv | ny_users.append(moscow_users).to_csv("./moscow_ny_bots.csv", encoding='utf8') | Twitter bots/Botnet search.ipynb | UserAd/data_science | mit |
Loading and visualizing the input data | # read the peaks
flt = columnfile('sma_261N.flt.new')
# peaks indexed to phase 1
phase1 = flt.copy()
phase1.filter( phase1.labels > -1 )
# unindexed peaks (phase 2 + unindexed phase 1?)
phase2 = flt.copy()
phase2.filter( phase2.labels == -1 )
#plot radial transform for phase 1
plt.plot( phase1.tth_per_grain, phase1.... | sandbox/weighted_kde/3DXRD diffractogram from filtered peaks.ipynb | jonwright/ImageD11 | gpl-2.0 |
Plotting the diffraction profile | # Probability density function (pdf) of 2theta
# weighted by the peak intensity and using default 2theta bandwidth
I_phase1 = phase1.sum_intensity * phase1.Lorentz_per_grain
pdf = wkde.gaussian_kde( phase1.tth_per_grain, weights = I_phase1)
# Plotting it over 2theta range
x = np.linspace( min(flt.tth), max(flt.tth), 5... | sandbox/weighted_kde/3DXRD diffractogram from filtered peaks.ipynb | jonwright/ImageD11 | gpl-2.0 |
The profile showed above is highly smoothed and the hkl peaks are merged.<br>
$\to$ A Smaller bandwidth should be used.
Choosing the right bandwidth of the estimator
The bandwidth can be passed as argument to the gaussian_kde() object or set afterward using the later set_badwidth() method. For example, the bandwidth ca... | pdf_phase1 = wkde.gaussian_kde( phase1.tth, weights = phase1.sum_intensity )
pdf_phase2 = wkde.gaussian_kde( phase2.tth, weights = phase2.sum_intensity )
frac_phase1 = np.sum( phase1.sum_intensity ) / np.sum( flt.sum_intensity )
frac_phase2 = np.sum( phase2.sum_intensity ) / np.sum( flt.sum_intensity )
from ipywidgets... | sandbox/weighted_kde/3DXRD diffractogram from filtered peaks.ipynb | jonwright/ImageD11 | gpl-2.0 |
Read some data | df1 = pd.read_csv('/Users/atma6951/Documents/code/pychakras/pychakras/udemy_ml_bootcamp/Python-for-Data-Visualization/Pandas Built-in Data Viz/df1', index_col=0)
df2 = pd.read_csv('/Users/atma6951/Documents/code/pychakras/pychakras/udemy_ml_bootcamp/Python-for-Data-Visualization/Pandas Built-in Data Viz/df2')
df1.head... | python_crash_course/pandas_data_viz_1.ipynb | AtmaMani/pyChakras | mit |
3 ways of calling plot from a DataFrame
df.plot() and specify the plot type, the X and Y columns etc
df.plot.hist() calling plot in OO fashion. Only specify teh X and Y and color or size columns
df['column'].plot.plotname() - calling plot on a series
Types of plot that can be called: area, bar, line, scatter, box, he... | df1.plot(x='A', kind='hist')
df1['A'].plot.hist(bins=30) | python_crash_course/pandas_data_viz_1.ipynb | AtmaMani/pyChakras | mit |
Plotting a histogram of all numeric columns in the dataframe: | df1.hist() | python_crash_course/pandas_data_viz_1.ipynb | AtmaMani/pyChakras | mit |
In reality, you have a lot more columns. You can prettify the above by creating a layout and figsize: | ax_list = df1.hist(bins=25, layout=(2,2), figsize=(7,7))
plt.tight_layout() | python_crash_course/pandas_data_viz_1.ipynb | AtmaMani/pyChakras | mit |
Plotting histogram of all columns and sharing axes
The chart above might make more sense if you shared the X as well as Y axes for different columns. This helps in comparing the distribution of values visually. | ax_list = df1.hist(bins=25, sharex=True, sharey=True, layout=(1,4), figsize=(15,4))
ax_list = df1.hist(bins=25, sharex=True, sharey=True, layout=(2,2), figsize=(8,8)) | python_crash_course/pandas_data_viz_1.ipynb | AtmaMani/pyChakras | mit |
Backgrounds
You can specify dark or white background and style info to the matplotlib that is used behind the scenes.
Area plot | plt.style.use('dark_background')
df2.plot.area() | python_crash_course/pandas_data_viz_1.ipynb | AtmaMani/pyChakras | mit |
Bar chart
Another style is fivethirtyeight | plt.style.use('fivethirtyeight')
df2.plot.bar() | python_crash_course/pandas_data_viz_1.ipynb | AtmaMani/pyChakras | mit |
Line plot
This is suited for time series data | #reset the style
plt.style.use('default')
# pass figsize to the matplotlib backend engine and `lw` is line width
df1.plot.line(x=df1.index, y='A', figsize=(12,2), lw=1) | python_crash_course/pandas_data_viz_1.ipynb | AtmaMani/pyChakras | mit |
Scatter plot
Use colormap or size to bring in a visualize a 3rd variable in your scatter | df1.plot.scatter(x='A', y='B',c='C', cmap='coolwarm')
# you could specify size s='c' however the points come out tiny.
# had to scale it by 100, hence using actual series data and not the column name
df2.plot.scatter(x='a',y='b', s=df2['c']*100) | python_crash_course/pandas_data_viz_1.ipynb | AtmaMani/pyChakras | mit |
KDE plots
To visualize the density of data | df1['A'].plot.kde() | python_crash_course/pandas_data_viz_1.ipynb | AtmaMani/pyChakras | mit |
Visualize the density of all columns in one plot | df1.plot.kde()
df2.plot.density() #I think density is an alias to KDE | python_crash_course/pandas_data_viz_1.ipynb | AtmaMani/pyChakras | mit |
Making wordclouds from text fields
Word clouds are a great way to visualize frequency of certain terms that appear in the data set. This is accomplished using the library wordcloud. You can install it as
conda install -c conda-forge wordcloud | registrant_df = pd.read_csv('./registrant.csv')
registrant_df.head() | python_crash_course/pandas_data_viz_1.ipynb | AtmaMani/pyChakras | mit |
Now, let us plot the responses from the column What would you like to learn? as a word cloud. First, we need to turn the series into a paragraph. | obj_series = registrant_df['What would you like to learn?'].dropna()
obj_list = list(obj_series)
obj_string = ' '.join(obj_list)
obj_string
from wordcloud import WordCloud
wc = WordCloud(width=1000, height=600, background_color='white')
obj_wc_img = wc.generate_from_text(obj_string)
plt.figure(figsize=(20,10))
plt.i... | python_crash_course/pandas_data_viz_1.ipynb | AtmaMani/pyChakras | mit |
Create a DataFrame object
Creat DataFrame by reading a file | mtcars = spark.read.csv(path='../../data/mtcars.csv',
sep=',',
encoding='UTF-8',
comment=None,
header=True,
inferSchema=True)
mtcars.show(n=5, truncate=False) | notebooks/01-data-strcture/1.2-dataframe.ipynb | MingChen0919/learning-apache-spark | mit |
Actuators
A multiprocessing block may need to interact asynchronously with some external device. To do so, the block puts data into a queue and uses threads responsible for interfacing between the queue and the device. This simple example illustrates the simplest actuator: a printer. Indeed printing can be done synchro... | import threading
from IoTPy.agent_types.sink import stream_to_queue
def f(in_streams, out_streams):
map_element(lambda v: v+100, in_streams[0], out_streams[0])
def source_thread_target(procs):
for i in range(3):
extend_stream(procs, data=list(range(i*2, (i+1)*2)), stream_name='x')
time.sleep(0... | examples/ExamplesOfMulticorePartTwo.ipynb | AssembleSoftware/IoTPy | bsd-3-clause |
Example of Process Structure with Feedback
The example shows a process structure with feedback. This example creates an echo from a spoken sound. (You can write more efficient and succinct code to compute echoes. The code in this example is here merely because it illustrates a concept.)
<br>
Streams
<ol>
<li><b>sou... | from IoTPy.agent_types.basics import *
def example_echo_two_cores():
# This is the delay from when the made sound hits a
# reflecting surface.
delay = 4
# This is the attenuation of the reflected wave.
attenuation = 0.5
# The results are put in this queue. A thread reads this
# queue and ... | examples/ExamplesOfMulticorePartTwo.ipynb | AssembleSoftware/IoTPy | bsd-3-clause |
Example source and actuator thread with single process
This example is the same as the previous one except that the computation is carried out in a single process rather in two processes. The example illustrates an actuator thread and a source thread in the same process. | def example_echo_single_core():
# This is the delay from when the made sound hits a
# reflecting surface.
delay = 4
# This is the attenuation of the reflected wave.
attenuation = 0.5
# The results are put in this queue. A thread reads this
# queue and feeds a speaker or headphone.
q = ... | examples/ExamplesOfMulticorePartTwo.ipynb | AssembleSoftware/IoTPy | bsd-3-clause |
Example of a grid computation
Grid computations are used in science, for example in computing the temperature of a metal plate. The grid is partitioned into regions with a process assigned to simulate each region. On the n-th step, each process reads the values of relevant parts of the grid and updates its own value.
<... | from IoTPy.core.stream import _no_value
def test_grid():
# N is the size of the grid
N = 5
# M is the number of steps of execution.
M = 5
# DELTA is the deviation from the final solution.
DELTA = 0.01
# even, odd are the grids that will be returned
# by this computation
even = multip... | examples/ExamplesOfMulticorePartTwo.ipynb | AssembleSoftware/IoTPy | bsd-3-clause |
let us avaluate a function of 3 variables on relatively large mesh | T = 1.618033988749895
from numpy import sin,cos,pi
r = 4.77
zmin,zmax = -r,r
xmin,xmax = -r,r
ymin,ymax = -r,r
Nx,Ny,Nz = 80,80,80
x = np.linspace(xmin,xmax,Nx)
y = np.linspace(ymin,ymax,Ny)
z = np.linspace(zmin,zmax,Nz)
x,y,z = np.meshgrid(x,y,z,indexing='ij')
%time p = 2 - (cos(x + T*y) + cos(x - T*y) + cos(y + T*z... | examples/SageDays74/implicit_plot3d_interactive.ipynb | K3D-tools/K3D-jupyter | mit |
isolevel can be changed from Python side: | p3d_1.level=-0.1
from ipywidgets import interact, interactive, fixed
import ipywidgets as widgets
@interact(l=widgets.FloatSlider(value=-.1,min=-3,max=1.1))
def g(l):
p3d_1.level=-l
| examples/SageDays74/implicit_plot3d_interactive.ipynb | K3D-tools/K3D-jupyter | mit |
to avoid recentering one can disable camera auto fit: | plot.camera_auto_fit = False
plot.grid_auto_fit = False | examples/SageDays74/implicit_plot3d_interactive.ipynb | K3D-tools/K3D-jupyter | mit |
one can add other plots to the same scene: | %%time
p =(x**2+y**2+z**2+2*y-1)*((x**2+y**2+z**2-2*y-1)**2-8*z**2)+16*x*z*(x**2+y**2+z**2-2*y-1)
plot += k3d.marching_cubes(p,xmin=xmin,xmax=xmax,ymin=ymin,ymax=ymax, zmin=zmin, zmax=zmax, level=0.0,color=0xff0000)
%%time
p = x**2 + y**2 - z**2 -0.
plot += k3d.marching_cubes(p,xmin=xmin,xmax=xmax,ymin=ymin,ymax=ymax... | examples/SageDays74/implicit_plot3d_interactive.ipynb | K3D-tools/K3D-jupyter | mit |
Config | import nlpaug.augmenter.char as nac
import nlpaug.augmenter.word as naw
import nlpaug.augmenter.sentence as nas
import nlpaug.flow as naf
from nlpaug.util import Action
text = 'The quick brown fox jumps over the lazy dog .'
print(text) | example/flow.ipynb | makcedward/nlpaug | mit |
Flow <a class="anchor" id="flow">
To make use of multiple augmentation, sequential and sometimes pipelines are introduced to connect augmenters.
Sequential Pipeline<a class="anchor" id="seq_pipeline">
Apply different augmenters sequentially | aug = naf.Sequential([
nac.RandomCharAug(action="insert"),
naw.RandomWordAug()
])
aug.augment(text) | example/flow.ipynb | makcedward/nlpaug | mit |
Generate mulitple synthetic data | aug = naf.Sequential([
nac.RandomCharAug(action="insert"),
naw.RandomWordAug()
])
aug.augment(text, n=3) | example/flow.ipynb | makcedward/nlpaug | mit |
Sometimes Pipeline<a class="anchor" id="sometimes_pipeline">
Apply some augmenters randomly | aug = naf.Sometimes([
nac.RandomCharAug(action="delete"),
nac.RandomCharAug(action="insert"),
naw.RandomWordAug()
])
aug.augment(text) | example/flow.ipynb | makcedward/nlpaug | mit |
$$
v(s) = \min_x(cost(x,s) + v(new state(x,s)))
$$ | class DynamicProgram(object):
"""
Generate a dynamic program to find a set of optimal descissions using the HJB.
define the program by:
Setting intial states via: set_inital_state(list or int)
Setting the number of steps via: set_step_number(int)
Add a set of desc... | Dynamic programming class.ipynb | icfly2/hjb_solvers | gpl-3.0 |
We can also solve classic dynamic programming problems such as the knapsack problem, hannoi towers or the fibonacci number calculation. Blank functions are outlined below.
The functions must fulfill a range of conditions:
$$
f:\mathcal{S}^n\times x \rightarrow \mathbb{R}
$$
where $\mathcal{S}$ is the set of permissab... | #help(DynamicProgram)
def cost(x,s):
"""Return a float or integer"""
pass
def new_state(x,s):
"""Return a tuple"""
pass
def val_T(s,settings):
"""Return a float or int"""
pass | Dynamic programming class.ipynb | icfly2/hjb_solvers | gpl-3.0 |
We can solve a very simple pump optimsiation where the state of water in a tank is given by h and described by:
$$
s_{new} = \begin{cases} (t+1,h-1) & \text{if } x = 0 \ (t+1,h+1) & \text{if } x = 1 \ (t+1,h+1.5) & \text{if } x = 2\end{cases}
$$
The operating cost are described by:
$$
cost = tarrif(t)\times x
$$
where... | def simple_cost(x,s):
tariff = [19, 8, 20, 3, 12, 14, 0, 4, 3, 13, 11, 13, 13, 11, 16, 14, 16,
19, 1, 8, 0, 4, 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,
12, 3, 18, 15, 3, 10, 12, 6, 3, 5, 11, 0, 11, 8, 10, 11, 5,
15, 8, 2, 0., 0., 0., 0., 0., 0.... | Dynamic programming class.ipynb | icfly2/hjb_solvers | gpl-3.0 |
We can have more than one state variable. For example we can add a second tank and now pump to either of them:
Here they have similar equations, but they can be completly independant, of each other. Any cost and state function that meets the requirments above is allowed.
$$
s_{new} = \begin{cases} (t+1,h_1-1,h_2-1) & \... | def simple_state2(x,s):
if x == 0:
return (s[0]+1,s[1]-1,s[2]-1)
elif x == 1:
return (s[0]+1,s[1]+1,s[2])
elif x == 2:
return (s[0]+1,s[1] ,s[2]+2)
# We also need to update the final value function.
def val_T2(s,settings):
if s[1] < settings['Initial state'][1] or s[2] < settin... | Dynamic programming class.ipynb | icfly2/hjb_solvers | gpl-3.0 |
One final example is the checkerboard problem as outlined here: https://en.wikipedia.org/wiki/Dynamic_programming#Checkerboard | board = np.array([[1,3,4,6],
[2,6,2,1],
[7,3,2,1],
[0,4,2,9]])
def cost(x,s):
"""Return a float or integer"""
global board
return board[s[0],s[1]]
def new_state(x,s):
"""Return a tuple"""
global board
return (int(s[0]+1),int(s[1]+x))
... | Dynamic programming class.ipynb | icfly2/hjb_solvers | gpl-3.0 |
An example we can solve is the water allocation problem from the tutorial sheets:
Consider a water supply allocation problem. Suppose that a quantity Q can be
allocated to three water users (indices j=1, 2 and 3), what is allocation x4 which
maximises the total net benefits?
The gross benefit resulting from the allocat... | Costs = np.array([[100,0.1, 10,0.6],
[50, 0.4, 10,0.8],
[100,0.2, 25,0.4]])
def value(x,s):
global Costs
return Costs[s[0],0]*(1-math.exp(-Costs[s[0],1]*x))
def cost(x,s):
"""Return a float or integer"""
global Costs
if
return ... | Dynamic programming class.ipynb | icfly2/hjb_solvers | gpl-3.0 |
Stochastic Programming
$$
v(s,i) = \min_x( cost(x,s,i)+\sum_j (p_{i,j} \times v(newstate(x,s,j))) )
$$
Where the probability $p_{i,j}$ is the probaility of jumping from state $i$ to state $j$. Currently the transition matrix is invariate, however this can be easily implimented with P as a list of lists. | class StochasticProgram(DynamicProgram):
"""
Adds a stochastic component to the dynamic program.
state now is: s where s[0] is the step s[1:-1] is the states of the system and s[-1] is the stochastic state
The transition matrix for the markov chain describing the stochastic bhavior is added by... | Dynamic programming class.ipynb | icfly2/hjb_solvers | gpl-3.0 |
The cost of operating a pump with a given wind turbine power input given by a certain state is given by:
$$
cost(x,t,h,j) := \begin{cases} T(t) \times (x \times P_p - W(t,j)) & \text{if} +ve \
E_{xp} \times (x \times P_p - W(t,j)) & \text{if} -ve\end{cases}
$$
where $W(t,j)$ is the wind power output at time $t$ with a... | # Convention s = t,h,j
def stoch_simple_state(x,s):
#print s
if x == 0:
return (s[0]+1,s[1]-1,s[2])
elif x == 1:
return (s[0]+1,s[1]+1,s[2])
elif x == 2:
return (s[0]+1,s[1]+1.5,s[2])
def err_corr_wind_power_cost(x,s):
Tariff = [5,5,5,5,5,8,8,8,8,8,12,12,12,12,1... | Dynamic programming class.ipynb | icfly2/hjb_solvers | gpl-3.0 |
The cost of operating a pump with a given wind turbine power input is given by:
$$
cost(x,t,h) := \begin{cases} T(t) \times (x \times P_p - W(t)) & \text{if} +ve \
E_{xp} \times (x \times P_p - W(t)) & \text{if} -ve\end{cases}
$$
where $x$ is the descision variable, $W(t)$ is the wind turbine output in time step $t$. ... | def wind_power_cost(x,s):
"""Very simple cost function for a pump with wind turbine power"""
Tariff = [5,5,5,5,5,8,8,8,8,8,12,12,12,12,12,50,50,50,50,20,20,6,5,5]
Wind = [46, 1, 3, 36, 30, 19, 9, 26, 35, 5, 49, 3, 6, 36, 43, 36, 14,
34, 2, 0, 0, 30, 13, 36]
Export_price = 5.5
... | Dynamic programming class.ipynb | icfly2/hjb_solvers | gpl-3.0 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.