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 |
|---|---|---|---|---|---|
**WOW. That's nice code!**Also: **Naming your baby Daenerys after the hero...**...is a bad break. | (baby_names
.query('name in ["Khaleesi","Ramsay","Lyanna","Ellaria","Meera"] & \
year >= 2000')
.groupby(['name','year'])['count'].sum() # for each name-year, combine M and F counts
.reset_index() # give use the column names back as they were (makes the plot call easy)
.pipe((sns.linepl... | _____no_output_____ | MIT | content/03/02f_chains-Copy1.ipynb | schemesmith/ledatascifi-2021 |
**BUT IT COULD BE WORSE** | (baby_names
.query('name in ["Krymson"] & year >= 1950')
.groupby(['name','year'])['count'].sum() # for each name-year, combine M and F counts
.reset_index() # give use the column names back as they were (makes the plot call easy)
.pipe((sns.lineplot, 'data'),hue='name',x='year',y='count')
) ... | _____no_output_____ | MIT | content/03/02f_chains-Copy1.ipynb | schemesmith/ledatascifi-2021 |
"Bookrating (Collaborative-Filtering)"> "Prediction of tangible books to read using collaborative filtering"- toc: false- branch: master- badges: true- comments: true- categories: [jupyter, pytorch, pytorch-lightning]- hide: false- search_exclude: true | %%capture
!pip install -U fastai
from google.colab import drive
drive.mount('/content/drive')
from fastai.collab import *
import pandas as pd
import torch.nn as nn
pathr = '/content/drive/MyDrive/my-datasets/collaborative-filtering/BX-Book-Ratings.csv'
pathb = '/content/drive/MyDrive/my-datasets/collaborative-filtering... | Harry Potter and the Goblet of Fire (Book 4)
Harry Potter and the Chamber of Secrets (Book 2)
The X-Planes: X-1 to X-45: 3rd Edition
Sanctuary: Finding Moments of Refuge in the Presence of God
Harry Potter and the Sorcerer's Stone (Book 1)
| Apache-2.0 | _notebooks/2021-01-26-Bookrating.ipynb | rajivreddy219/ai-projects |
Text classification with attention and synthetic gradients.Imports and set-up: | %tensorflow_version 2.x
import numpy as np
import tensorflow as tf
import pandas as pd
import subprocess
from sklearn.model_selection import train_test_split
import gensim
import re
import sys
import time
# TODO: actually implement distribution in the training loop
strategy = tf.distribute.get_strategy()
use_mixed_pr... | _____no_output_____ | CC0-1.0 | synth_grads.ipynb | tarcey/Synthetic-Gradients-in-TF |
Downloading the data | # Download the Sentiment140 dataset
!mkdir -p data
!wget -nc https://nyc3.digitaloceanspaces.com/ml-files-distro/v1/sentiment-analysis-is-bad/data/training.1600000.processed.noemoticon.csv.zip -P data
!unzip -n -d data data/training.1600000.processed.noemoticon.csv.zip | _____no_output_____ | CC0-1.0 | synth_grads.ipynb | tarcey/Synthetic-Gradients-in-TF |
Loading and splitting the data | sen140 = pd.read_csv(
"data/training.1600000.processed.noemoticon.csv", encoding='latin-1',
names=["target", "ids", "date", "flag", "user", "text"])
sen140.head()
sen140 = sen140.sample(frac=1).reset_index(drop=True)
sen140 = sen140[['text', 'target']]
features, targets = sen140.iloc[:, 0].values, sen140.iloc[:... | _____no_output_____ | CC0-1.0 | synth_grads.ipynb | tarcey/Synthetic-Gradients-in-TF |
Preprocessing data | def process_tweet(x):
x = x.strip()
x = x.lower()
x = re.sub(r"[^a-zA-Z0-9üöäÜÖÄß\.,!\?\-%\$€\/ ]+'", ' ', x)
x = re.sub('([\.,!\?\-%\$€\/])', r' \1 ', x)
x = re.sub('\s{2,}', ' ', x)
x = x.split()
x.append("[&END&]")
length = len(x)
return x
tweets_train = []
tweets_test = []
for ... | _____no_output_____ | CC0-1.0 | synth_grads.ipynb | tarcey/Synthetic-Gradients-in-TF |
Using gensim word2vec to get a good word embedding. | # train the embedding
embedding_dims = 128
embedding = gensim.models.Word2Vec(tweets_train,
size=embedding_dims, min_count=0)
def tokenize(_in, _vocab):
_out = []
for i in range(len(_in)):
tweet = _in[i]
wordlist = []
for word in tweet:
word... | _____no_output_____ | CC0-1.0 | synth_grads.ipynb | tarcey/Synthetic-Gradients-in-TF |
Creating modules and defining the model. | class SequenceCollapseAttention(tf.Module):
'''
Collapses a sequence of arbitrary length into num_out_entries entries from
the sequence according to dot-product attention. So, a variable length
sequence is reduced to a sequence of a fixed, known length.
'''
def __init__(self,
... | _____no_output_____ | CC0-1.0 | synth_grads.ipynb | tarcey/Synthetic-Gradients-in-TF |
Training the model | def train_validation_loop(model_caller, data_generator, epochs, metrics=[]):
batch_time = -1
for epoch in range(epochs):
start_e = time.time()
start_p = time.time()
num_batches = len(data_generator)
predictions = [None] * num_batches
for b in range(num_batches):
... | _____no_output_____ | CC0-1.0 | synth_grads.ipynb | tarcey/Synthetic-Gradients-in-TF |
Testing it on validation data | def validator(model):
@tf.function(experimental_relax_shapes=True)
def validation_step(x_batch, y_batch, model=model, metrics=[]):
predictions = model(x_batch, training=False)
for metric in metrics:
metric.update_state(y_batch, predictions)
return predictions
return vali... | _____no_output_____ | CC0-1.0 | synth_grads.ipynb | tarcey/Synthetic-Gradients-in-TF |
Get some example results from the the test data. | most_evil_tweet=None
most_evil_evilness=1
most_cool_tweet=None
most_cool_coolness=1
most_angelic_tweet=None
most_angelic_angelicness=0
y_pred = np.concatenate(predictions)
for i in range(0,len(y_pred)):
judgement = y_pred[i]
polarity = abs(judgement-0.5)*2
if judgement>=most_angelic_angelicness:
mo... | _____no_output_____ | CC0-1.0 | synth_grads.ipynb | tarcey/Synthetic-Gradients-in-TF |
Imports | import numpy as np
import xarray as xr
import seaborn as sns
import matplotlib as mpl
import matplotlib.pyplot as plt
from matplotlib.colors import LinearSegmentedColormap
from matplotlib.patches import Polygon
from matplotlib import colors as mat_colors
import mpl_toolkits.axisartist as axisartist
from mpl_toolkits.ax... | _____no_output_____ | BSD-3-Clause | Fig.3.11_par_methods_overview/Par_method_comparision_overview.ipynb | pat-schmitt/supplementary_material_master_thesis |
Define Functions Performance measurements | def BIAS(a1, a2):
return (a1 - a2).mean().item()
def RMSE(a1, a2):
return np.sqrt(((a1 - a2)**2).mean()).item()
def DIFF(a1, a2):
return np.max(np.abs(a1 - a2)).item() | _____no_output_____ | BSD-3-Clause | Fig.3.11_par_methods_overview/Par_method_comparision_overview.ipynb | pat-schmitt/supplementary_material_master_thesis |
help function for heatmap axis | def setup_axes(fig, rect):
ax = axisartist.Subplot(fig, rect)
fig.add_subplot(ax)
return ax | _____no_output_____ | BSD-3-Clause | Fig.3.11_par_methods_overview/Par_method_comparision_overview.ipynb | pat-schmitt/supplementary_material_master_thesis |
heatmap | def heatmap(datasets, # first_dataset, second_dataset,
opti_var,
annotation=None,
annotation_x_position=0,
annotation_y_position=1,
fig=None, ax=None,
cmap='vlag',
cmap_levels=None,
grid_color='grey',
grid_linewi... | _____no_output_____ | BSD-3-Clause | Fig.3.11_par_methods_overview/Par_method_comparision_overview.ipynb | pat-schmitt/supplementary_material_master_thesis |
legend plot | def add_legend2(ax,
title,
fontsize,
lw,
ms,
labels):
ax.plot([],
[],
'-',
lw=lw,
ms=ms,
c='none',
label=labels[0])
# plot for first gradient scaling
a... | _____no_output_____ | BSD-3-Clause | Fig.3.11_par_methods_overview/Par_method_comparision_overview.ipynb | pat-schmitt/supplementary_material_master_thesis |
performance plot | def performance_plot(ax,
datasets,
fig=None,
# 'bed_h RMSE', 'bed_h Diff', 'bed_h Bias',
# 'bed_shape RMSE', 'bed_shape Diff', 'bed_shape Bias',
# 'w0 RMSE', 'w0 Diff', 'w0 Bias',
# 'sfc_h RMSE'... | _____no_output_____ | BSD-3-Clause | Fig.3.11_par_methods_overview/Par_method_comparision_overview.ipynb | pat-schmitt/supplementary_material_master_thesis |
Define Colors | colors = sns.color_palette("colorblind")
colors
axis_color = list(colors[7]) + [1.]
color_1 = list(colors[3]) + [1.]
color_2 = list(colors[0]) + [1.]
color_3 = list(colors[4]) + [1.]
color_4 = list(colors[2]) + [1.]
glacier_color = glacier_color = list(colors[9]) + [.5] | _____no_output_____ | BSD-3-Clause | Fig.3.11_par_methods_overview/Par_method_comparision_overview.ipynb | pat-schmitt/supplementary_material_master_thesis |
Import Data | input_folder = 'plot_data/'
filename_scale_1 = 'par_clif_cons_ret_bed_h_and_bed_shape_at_once_scal1reg11.nc'
filename_scale_1e4 = 'par_clif_cons_ret_bed_h_and_bed_shape_at_once_scal1e4reg11.nc'
filename_separated = 'par_clif_cons_ret_bed_h_and_bed_shape_separatedreg11.nc'
filename_calculated = 'par_clif_cons_ret_bed_h_... | _____no_output_____ | BSD-3-Clause | Fig.3.11_par_methods_overview/Par_method_comparision_overview.ipynb | pat-schmitt/supplementary_material_master_thesis |
Create figure with performance | #dataset,
nr_of_iterations = None
facecolor = 'white'
labels_pad = -700
cmap = 'Spectral'
fontsize = 25
lw=2
ms=10
annotation_x_position_spatial = -0.15
annotation_y_position_spatial = 0.9
annotation_x_position_performance = -0.14
annotation_y_position_performance = 1.05
#index_start_first_profil_row = 0
#index_end_fir... | _____no_output_____ | BSD-3-Clause | Fig.3.11_par_methods_overview/Par_method_comparision_overview.ipynb | pat-schmitt/supplementary_material_master_thesis |
Prune DB | conn = sqlite3.connect('prices.db')
conn.executescript(SCHEMA)
conn.commit()
conn.close() | _____no_output_____ | MIT | crypto-examples/Insert in Database.ipynb | zzhengnan/pycon-concurrency-tutorial-2020 |
Insert Data | BASE_PATH = Path('crypto_data/')
files = list(BASE_PATH.glob('*.csv'))
INSERT_STATEMENT = """
INSERT INTO price (
exchange, symbol, open, high, low, close, volume, day
) VALUES (?, ?, ?, ?, ?, ?, ?, ?);
"""
conn = sqlite3.connect('prices.db')
for file in files:
exchange, symbol = file.name[:-4].split('_')
d... | _____no_output_____ | MIT | crypto-examples/Insert in Database.ipynb | zzhengnan/pycon-concurrency-tutorial-2020 |
Final Test | conn = sqlite3.connect('prices.db')
cursor = conn.cursor()
cursor.execute('SELECT COUNT(*) FROM price;')
cursor.fetchone()
cursor.execute('SELECT * FROM price LIMIT 5;')
cursor.fetchall()
conn.close() | _____no_output_____ | MIT | crypto-examples/Insert in Database.ipynb | zzhengnan/pycon-concurrency-tutorial-2020 |
Exchanges | conn = sqlite3.connect('prices.db')
cursor = conn.cursor()
cursor.execute('SELECT DISTINCT exchange FROM price;')
cursor.fetchall()
cursor.execute('SELECT DISTINCT symbol FROM price;')
cursor.fetchall() | _____no_output_____ | MIT | crypto-examples/Insert in Database.ipynb | zzhengnan/pycon-concurrency-tutorial-2020 |
Filtered query: | cursor.execute('SELECT * FROM price WHERE symbol = "btc" AND exchange = "bitfinex" AND day = "2019-07-20";')
cursor.fetchall() | _____no_output_____ | MIT | crypto-examples/Insert in Database.ipynb | zzhengnan/pycon-concurrency-tutorial-2020 |
Setup Data | #----------Training data set
df_train0 = pd.read_json('../input/deepfake/metadata0.json')
df_train1 = pd.read_json('../input/deepfake/metadata1.json')
df_train2 = pd.read_json('../input/deepfake/metadata2.json')
df_train3 = pd.read_json('../input/deepfake/metadata3.json')
df_train4 = pd.read_json('../input/deepfake/met... | _____no_output_____ | MIT | DFDC/Training Xception Classifier .ipynb | nizamphoenix/kaggle |
Dataset | def read_img(path):
return cv2.cvtColor(cv2.imread(path),cv2.COLOR_BGR2RGB)
import random
def shuffle(X,y):
new = []
for m,n in zip(X,y):
new.append([m,n])
random.shuffle(new)
X,y = [],[]
for path,label in new:
X.append(path)
y.append(label)
return X,y | _____no_output_____ | MIT | DFDC/Training Xception Classifier .ipynb | nizamphoenix/kaggle |
FAKE-->1 REAL-->0 | def get_data(train_paths, train_y, val_paths, val_y):
train_X=[]
for img in tqdm(train_paths):
train_X.append(read_img(img))
val_X=[]
for img in tqdm(val_paths):
val_X.append(read_img(img))
train_X, train_y = shuffle(train_X,train_y)
val_X, val_y = shuffle(val_X,val_y)
return... | _____no_output_____ | MIT | DFDC/Training Xception Classifier .ipynb | nizamphoenix/kaggle |
Model | !pip install pytorchcv --quiet
from pytorchcv.model_provider import get_model
model = get_model("xception", pretrained=True)
# model = get_model("resnet18", pretrained=True)
model = nn.Sequential(*list(model.children())[:-1]) # Remove original output layer
model[0].final_block.pool = nn.Sequential(nn.AdaptiveAvgPool2d(... | _____no_output_____ | MIT | DFDC/Training Xception Classifier .ipynb | nizamphoenix/kaggle |
Train Functions | def calculate_loss(preds, targets):
return F.binary_cross_entropy(F.sigmoid(preds), targets)
def train_model(epoch, optimizer, scheduler=None, history=None):
model.train()
total_loss = 0
t = tqdm(train_loader)
for i, (img_batch, y_batch) in enumerate(t):
img_batch = img_batch.cuda().fl... | _____no_output_____ | MIT | DFDC/Training Xception Classifier .ipynb | nizamphoenix/kaggle |
Dataloaders | X, val_X, y, val_y = get_data(train_img_paths, train_img_labels,val_img_paths, val_img_labels)
print('There are '+str(y.count(1))+' fake train samples')
print('There are '+str(y.count(0))+' real train samples')
print('There are '+str(val_y.count(1))+' fake val samples')
print('There are '+str(val_y.count(0))+' real va... | _____no_output_____ | MIT | DFDC/Training Xception Classifier .ipynb | nizamphoenix/kaggle |
Train | import gc
history = pd.DataFrame()
history2 = pd.DataFrame()
torch.cuda.empty_cache()
gc.collect()
best = 1e10
n_epochs = 20
batch_size = 64#BATCH SIZE CHANGED
train_loader = DataLoader(dataset=train_dataset, batch_size=batch_size, shuffle=True, num_workers=4)
val_loader = DataLoader(dataset=val_dataset, batch_size... | _____no_output_____ | MIT | DFDC/Training Xception Classifier .ipynb | nizamphoenix/kaggle |
Face RecognitionIn this assignment, you will build a face recognition system. Many of the ideas presented here are from [FaceNet](https://arxiv.org/pdf/1503.03832.pdf). In lecture, we also talked about [DeepFace](https://research.fb.com/wp-content/uploads/2016/11/deepface-closing-the-gap-to-human-level-performance-in-... | from keras.models import Sequential
from keras.layers import Conv2D, ZeroPadding2D, Activation, Input, concatenate
from keras.models import Model
from keras.layers.normalization import BatchNormalization
from keras.layers.pooling import MaxPooling2D, AveragePooling2D
from keras.layers.merge import Concatenate
from kera... | Using TensorFlow backend.
| MIT | 4. Convolutional Neural Networks/Face_Recognition_v3a.ipynb | taktak-hi/Deep-Learning-Specialization |
0 - Naive Face VerificationIn Face Verification, you're given two images and you have to determine if they are of the same person. The simplest way to do this is to compare the two images pixel-by-pixel. If the distance between the raw images are less than a chosen threshold, it may be the same person! **Figure 1**... | FRmodel = faceRecoModel(input_shape=(3, 96, 96))
print("Total Params:", FRmodel.count_params()) | Total Params: 3743280
| MIT | 4. Convolutional Neural Networks/Face_Recognition_v3a.ipynb | taktak-hi/Deep-Learning-Specialization |
** Expected Output **Total Params: 3743280 By using a 128-neuron fully connected layer as its last layer, the model ensures that the output is an encoding vector of size 128. You then use the encodings to compare two face images as follows: **Figure 2**: By computing the distance between two encodings and threshol... | # GRADED FUNCTION: triplet_loss
def triplet_loss(y_true, y_pred, alpha = 0.2):
"""
Implementation of the triplet loss as defined by formula (3)
Arguments:
y_true -- true labels, required when you define a loss in Keras, you don't need it in this function.
y_pred -- python list containing three... | loss = 528.143
| MIT | 4. Convolutional Neural Networks/Face_Recognition_v3a.ipynb | taktak-hi/Deep-Learning-Specialization |
**Expected Output**: **loss** 528.143 2 - Loading the pre-trained modelFaceNet is trained by minimizing the triplet loss. But since training requires a lot of data and a lot of computation, we won't train it from scratch here. Instead, we load a previously t... | FRmodel.compile(optimizer = 'adam', loss = triplet_loss, metrics = ['accuracy'])
load_weights_from_FaceNet(FRmodel) | _____no_output_____ | MIT | 4. Convolutional Neural Networks/Face_Recognition_v3a.ipynb | taktak-hi/Deep-Learning-Specialization |
Here are some examples of distances between the encodings between three individuals: **Figure 4**: Example of distance outputs between three individuals' encodingsLet's now use this model to perform face verification and face recognition! 3 - Applying the model You are building a system for an office building wh... | database = {}
database["danielle"] = img_to_encoding("images/danielle.png", FRmodel)
database["younes"] = img_to_encoding("images/younes.jpg", FRmodel)
database["tian"] = img_to_encoding("images/tian.jpg", FRmodel)
database["andrew"] = img_to_encoding("images/andrew.jpg", FRmodel)
database["kian"] = img_to_encoding("im... | _____no_output_____ | MIT | 4. Convolutional Neural Networks/Face_Recognition_v3a.ipynb | taktak-hi/Deep-Learning-Specialization |
Now, when someone shows up at your front door and swipes their ID card (thus giving you their name), you can look up their encoding in the database, and use it to check if the person standing at the front door matches the name on the ID.**Exercise**: Implement the verify() function which checks if the front-door camera... | # GRADED FUNCTION: verify
def verify(image_path, identity, database, model):
"""
Function that verifies if the person on the "image_path" image is "identity".
Arguments:
image_path -- path to an image
identity -- string, name of the person you'd like to verify the identity. Has to be an employ... | _____no_output_____ | MIT | 4. Convolutional Neural Networks/Face_Recognition_v3a.ipynb | taktak-hi/Deep-Learning-Specialization |
Younes is trying to enter the office and the camera takes a picture of him ("images/camera_0.jpg"). Let's run your verification algorithm on this picture: | verify("images/camera_0.jpg", "younes", database, FRmodel) | It's younes, welcome in!
| MIT | 4. Convolutional Neural Networks/Face_Recognition_v3a.ipynb | taktak-hi/Deep-Learning-Specialization |
**Expected Output**: **It's younes, welcome in!** (0.65939283, True) Benoit, who does not work in the office, stole Kian's ID card and tried to enter the office. The camera took a picture of Benoit ("images/camera_2.jpg). Let's run the verification algorithm ... | verify("images/camera_2.jpg", "kian", database, FRmodel) | It's not kian, please go away
| MIT | 4. Convolutional Neural Networks/Face_Recognition_v3a.ipynb | taktak-hi/Deep-Learning-Specialization |
**Expected Output**: **It's not kian, please go away** (0.86224014, False) 3.2 - Face RecognitionYour face verification system is mostly working well. But since Kian got his ID card stolen, when he came back to the office the next day and couldn't get in! To... | # GRADED FUNCTION: who_is_it
def who_is_it(image_path, database, model):
"""
Implements face recognition for the office by finding who is the person on the image_path image.
Arguments:
image_path -- path to an image
database -- database containing image encodings along with the name of the per... | _____no_output_____ | MIT | 4. Convolutional Neural Networks/Face_Recognition_v3a.ipynb | taktak-hi/Deep-Learning-Specialization |
Younes is at the front-door and the camera takes a picture of him ("images/camera_0.jpg"). Let's see if your who_it_is() algorithm identifies Younes. | who_is_it("images/camera_0.jpg", database, FRmodel) | it's younes, the distance is 0.659393
| MIT | 4. Convolutional Neural Networks/Face_Recognition_v3a.ipynb | taktak-hi/Deep-Learning-Specialization |
Logistic Regression | model = LogisticRegressionWithSGD.train(trainingData)
posTest = tf.transform('O M G GET cheap stuff by sending money to ...'.split(' '))
negTest = tf.transform('Hi Dad, I started studying Spark the other ...'.split(' '))
print('Prediction for positive test example: %g' % model.predict(posTest))
print('Prediction for ne... | Prediction for positive test example: 0
Prediction for negative test example: 0
| Apache-2.0 | a01_PySpark/e01_Resources/PySpark-and-MLlib/learning_spark_MLlib.ipynb | mindis/Big_Data_Analysis |
Creating vectors | from numpy import array
from pyspark.mllib.linalg import Vectors
denseVec1 = array([1.0, 2.0, 3.0])
denseVec2 = Vectors.dense([1.0, 2.0, 3.0])
sparseVec1 = Vectors.sparse(4, {0:1.0, 2:2.0})
sparseVec2 = Vectors.sparse(4, [0, 2], [1.0, 2.0]) | _____no_output_____ | Apache-2.0 | a01_PySpark/e01_Resources/PySpark-and-MLlib/learning_spark_MLlib.ipynb | mindis/Big_Data_Analysis |
Using HashingTF | sentence = 'hello hello world'
words = sentence.split()
tf = HashingTF(10000)
tf.transform(words)
rdd = sc.wholeTextFiles("data").map(lambda (name, text): text.split())
tfVectors = tf.transform(rdd) | _____no_output_____ | Apache-2.0 | a01_PySpark/e01_Resources/PySpark-and-MLlib/learning_spark_MLlib.ipynb | mindis/Big_Data_Analysis |
Copyright 2019 Digital Advantage - Deep Insider. | #@title Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# https://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under... | _____no_output_____ | Apache-2.0 | docs/articles/deeplearningdat.ipynb | DeepInsider/playground-data |
連載『機械学習 ＆ ディープラーニング入門（データ構造編）』のノートブック Deep Insiderで記事を読む Google Colabで実行する GitHubでソースコードを見る ※上から順に実行してください。上のコードで実行したものを再利用しているところがあるため、すべて実行しないとエラーになるコードがあります。 すべてのコードを一括実行したい場合は、メニューバーから［ランタイム］－［すべてのセルを実行］をクリックしてください。 ※このノートブックは「Python 2」でも実行できるようにしていますが、基本的に「Python 3」を利用することをお勧めします。 Pytho... | # Python バージョン2への対応
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
from __future__ import unicode_literals
import sys
print(sys.version_info.major) # 3 # バージョン（メジャー）
print(sys.version_info.minor) # 6 # バージョン（マイナー） | _____no_output_____ | Apache-2.0 | docs/articles/deeplearningdat.ipynb | DeepInsider/playground-data |
Python言語におけるデータの構造 Pythonにおける「1つの」データの表現 リスト1-1 「単一の」データを表現するコード | height = 177.2
print(height) # 177.2と出力される | _____no_output_____ | Apache-2.0 | docs/articles/deeplearningdat.ipynb | DeepInsider/playground-data |
リスト1-2 変数名だけを記述してオブジェクトの評価結果を出力 | height # 177.2と出力される | _____no_output_____ | Apache-2.0 | docs/articles/deeplearningdat.ipynb | DeepInsider/playground-data |
リスト1-3 オブジェクト評価結果の出力とprint()関数の出力の違い | import numpy as np
array2d = np.array([ [ 165.5, 58.4 ],
[ 177.2, 67.8 ],
[ 183.2, 83.7 ] ])
print(array2d) # [[165.5 58.4]
# [177.2 67.8]
# [183.2 83.7]]
array2d # array([[165.5, 58.4],
# [177.2, 67.8]... | _____no_output_____ | Apache-2.0 | docs/articles/deeplearningdat.ipynb | DeepInsider/playground-data |
リスト2-1 「単一の」データを複数書いて表現するコード | hana_height = 165.5
taro_height = 177.2
jiro_height = 183.2
hana_height, taro_height, jiro_height # (165.5, 177.2, 183.2) | _____no_output_____ | Apache-2.0 | docs/articles/deeplearningdat.ipynb | DeepInsider/playground-data |
リスト2-2 「複数（1次元）の」データを表現するコード | heights = [ 165.5, 177.2, 183.2 ]
heights # [165.5, 177.2, 183.2] | _____no_output_____ | Apache-2.0 | docs/articles/deeplearningdat.ipynb | DeepInsider/playground-data |
Pythonにおける「複数（2次元）の」データの表現 リスト3 「複数（2次元）の」データを表現するコード | people = [ [ 165.5, 58.4 ],
[ 177.2, 67.8 ],
[ 183.2, 83.7 ] ]
people # [165.5, 177.2, 183.2] | _____no_output_____ | Apache-2.0 | docs/articles/deeplearningdat.ipynb | DeepInsider/playground-data |
Pythonにおける「複数（多次元）の」データの表現 リスト4 「複数（3次元）の」データを表現するコード | list3d = [
[ [ 165.5, 58.4 ], [ 177.2, 67.8 ], [ 183.2, 83.7 ] ],
[ [ 155.5, 48.4 ], [ 167.2, 57.8 ], [ 173.2, 73.7 ] ],
[ [ 145.5, 38.4 ], [ 157.2, 47.8 ], [ 163.2, 63.7 ] ]
]
list3d # [[[165.5, 58.4], [177.2, 67.8], [183.2, 83.7]],
# [[155.5, 48.4], [167.2, 57.8], [173.2, 73.7]],
# [[145.5, ... | _____no_output_____ | Apache-2.0 | docs/articles/deeplearningdat.ipynb | DeepInsider/playground-data |
AIプログラムにおけるデータの構造（基本編） NumPyのインストール リスト5-1 `numpy`パッケージをインストールするためのシェルコマンド | !pip install numpy | _____no_output_____ | Apache-2.0 | docs/articles/deeplearningdat.ipynb | DeepInsider/playground-data |
numpyモジュールのインポート リスト5-2 `numpy`モジュールをインポートするコード例 | import numpy as np | _____no_output_____ | Apache-2.0 | docs/articles/deeplearningdat.ipynb | DeepInsider/playground-data |
NumPyのデータ型「多次元配列」オブジェクトの作成 リスト5-3 `array`関数で多次元配列を作成するコード例（値を使用） | array2d = np.array([ [ 165.5, 58.4 ],
[ 177.2, 67.8 ],
[ 183.2, 83.7 ] ])
array2d # array([[165.5, 58.4],
# [177.2, 67.8],
# [183.2, 83.7]]) | _____no_output_____ | Apache-2.0 | docs/articles/deeplearningdat.ipynb | DeepInsider/playground-data |
リスト5-4 `array`関数で多次元配列を作成するコード例（変数を使用） | array3d = np.array(list3d)
array3d # array([[[165.5, 58.4],
# [177.2, 67.8],
# [183.2, 83.7]],
#
# [[155.5, 48.4],
# [167.2, 57.8],
# [173.2, 73.7]],
#
# [[145.5, 38.4],
# [1... | _____no_output_____ | Apache-2.0 | docs/articles/deeplearningdat.ipynb | DeepInsider/playground-data |
リスト5-5 `ndarray`クラスの`tolist()`メソッドで多次元リストに変換するコード例 | tolist3d = array3d.tolist()
tolist3d # [[[165.5, 58.4], [177.2, 67.8], [183.2, 83.7]],
# [[155.5, 48.4], [167.2, 57.8], [173.2, 73.7]],
# [[145.5, 38.4], [157.2, 47.8], [163.2, 63.7]]] | _____no_output_____ | Apache-2.0 | docs/articles/deeplearningdat.ipynb | DeepInsider/playground-data |
AIプログラムにおけるデータの構造（応用編） Pandasのインストール リスト6 ◎pandas◎パッケージをインストールするためのシェルコマンド | !pip install pandas | _____no_output_____ | Apache-2.0 | docs/articles/deeplearningdat.ipynb | DeepInsider/playground-data |
図7-1 NumPyのデータをPandasで一覧表として表示する例 | import pandas as pd
df = pd.DataFrame(array2d, columns=['身長', '体重'])
df | _____no_output_____ | Apache-2.0 | docs/articles/deeplearningdat.ipynb | DeepInsider/playground-data |
AIプログラムにおけるデータの計算 AI・ディープラーニングで数学を使う理由 リスト7-1 3人の身長の平均を計算するコード例（個別の値を使用） | # hana_height, taro_height, jiro_height = 165.5, 177.2, 183.2 # Lesson 1のリスト2-1で宣言済み
average_height = (
hana_height +
taro_height +
jiro_height
) / 3
print(average_height) # 175.29999999999998 | _____no_output_____ | Apache-2.0 | docs/articles/deeplearningdat.ipynb | DeepInsider/playground-data |
リスト7-2 3人の身長と体重の平均を計算するコード例（多次元配列を使用） | import numpy as np
array1d = np.array([ 165.5, 177.2, 183.2 ])
average_height = np.average(array1d)
average_height # 175.29999999999998 | _____no_output_____ | Apache-2.0 | docs/articles/deeplearningdat.ipynb | DeepInsider/playground-data |
NumPyを使った計算 リスト8-1 3行2列の行列のさまざまな特性を表示するコード例 | array2d = np.array([ [ 165.5, 58.4 ],
[ 177.2, 67.8 ],
[ 183.2, 83.7 ] ])
print(array2d.shape) # (3, 2)
print(array2d.ndim) # 2
print(array2d.size) # 6 | _____no_output_____ | Apache-2.0 | docs/articles/deeplearningdat.ipynb | DeepInsider/playground-data |
リスト8-2 NumPyを使った行列計算 | diet = np.array([ [ 1.0, 0.0 ],
[ 0.0, 0.9 ] ])
lose_weights = diet @ array2d.T
# Python 3.5以降の場合。それ以前のPython 2系などの場合は、以下のmatmul関数を使う必要がある
#lose_weights = np.matmul(diet, array2d.T)
print(lose_weights.T) # [[165.5 52.56]
# [177.2 61.02]
# [183.2 ... | _____no_output_____ | Apache-2.0 | docs/articles/deeplearningdat.ipynb | DeepInsider/playground-data |
リスト8-3 全要素の平均値を算出（身長／体重別ではない） | averages = np.average(array2d)
averages # 122.63333333333334 | _____no_output_____ | Apache-2.0 | docs/articles/deeplearningdat.ipynb | DeepInsider/playground-data |
リスト8-4 身長／体重別の平均値を算出 | averages = np.average(array2d, axis=0)
averages # array([175.3 , 69.96666667]) | _____no_output_____ | Apache-2.0 | docs/articles/deeplearningdat.ipynb | DeepInsider/playground-data |
リスト8-5 3次元配列データでグループごとの身長／体重別の平均値を算出 | array3d = np.array(
[ [ [ 165.5, 58.4 ], [ 177.2, 67.8 ], [ 183.2, 83.7 ] ],
[ [ 155.5, 48.4 ], [ 167.2, 57.8 ], [ 173.2, 73.7 ] ],
[ [ 145.5, 38.4 ], [ 157.2, 47.8 ], [ 163.2, 63.7 ] ] ]
)
avr3d = np.average(array3d, axis=1)
print(avr3d) # [[175.3 69.96666667]
# [165.3 59.96... | _____no_output_____ | Apache-2.0 | docs/articles/deeplearningdat.ipynb | DeepInsider/playground-data |
Copyright 2019 The TensorFlow Authors. | #@title Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# https://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under... | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
Time series forecasting View on TensorFlow.org Run in Google Colab View source on GitHub Download notebook This tutorial is an introduction to time series forecasting using TensorFlow. It builds a few different styles of models including Convolutional and Recurrent Neural Networks (CNNs an... | import os
import datetime
import IPython
import IPython.display
import matplotlib as mpl
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import seaborn as sns
import tensorflow as tf
mpl.rcParams['figure.figsize'] = (8, 6)
mpl.rcParams['axes.grid'] = False | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
The weather datasetThis tutorial uses a weather time series dataset recorded by the Max Planck Institute for Biogeochemistry.This dataset contains 14 different features such as air temperature, atmospheric pressure, and humidity. These were collected every 10 minutes, beginning in 2003. For efficiency, you will use on... | zip_path = tf.keras.utils.get_file(
origin='https://storage.googleapis.com/tensorflow/tf-keras-datasets/jena_climate_2009_2016.csv.zip',
fname='jena_climate_2009_2016.csv.zip',
extract=True)
csv_path, _ = os.path.splitext(zip_path) | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
This tutorial will just deal with **hourly predictions**, so start by sub-sampling the data from 10 minute intervals to 1h: | df = pd.read_csv(csv_path)
# slice [start:stop:step], starting from index 5 take every 6th record.
df = df[5::6]
date_time = pd.to_datetime(df.pop('Date Time'), format='%d.%m.%Y %H:%M:%S') | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
Let's take a glance at the data. Here are the first few rows: | df.head() | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
Here is the evolution of a few features over time. | plot_cols = ['T (degC)', 'p (mbar)', 'rho (g/m**3)']
plot_features = df[plot_cols]
plot_features.index = date_time
_ = plot_features.plot(subplots=True)
plot_features = df[plot_cols][:480]
plot_features.index = date_time[:480]
_ = plot_features.plot(subplots=True) | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
Inspect and cleanup Next look at the statistics of the dataset: | df.describe().transpose() | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
Wind velocity One thing that should stand out is the `min` value of the wind velocity, `wv (m/s)` and `max. wv (m/s)` columns. This `-9999` is likely erroneous. There's a separate wind direction column, so the velocity should be `>=0`. Replace it with zeros: | wv = df['wv (m/s)']
bad_wv = wv == -9999.0
wv[bad_wv] = 0.0
max_wv = df['max. wv (m/s)']
bad_max_wv = max_wv == -9999.0
max_wv[bad_max_wv] = 0.0
# The above inplace edits are reflected in the DataFrame
df['wv (m/s)'].min() | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
Feature engineeringBefore diving in to build a model it's important to understand your data, and be sure that you're passing the model appropriately formatted data. WindThe last column of the data, `wd (deg)`, gives the wind direction in units of degrees. Angles do not make good model inputs, 360° and 0° should be cl... | plt.hist2d(df['wd (deg)'], df['wv (m/s)'], bins=(50, 50), vmax=400)
plt.colorbar()
plt.xlabel('Wind Direction [deg]')
plt.ylabel('Wind Velocity [m/s]') | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
But this will be easier for the model to interpret if you convert the wind direction and velocity columns to a wind **vector**: | wv = df.pop('wv (m/s)')
max_wv = df.pop('max. wv (m/s)')
# Convert to radians.
wd_rad = df.pop('wd (deg)')*np.pi / 180
# Calculate the wind x and y components.
df['Wx'] = wv*np.cos(wd_rad)
df['Wy'] = wv*np.sin(wd_rad)
# Calculate the max wind x and y components.
df['max Wx'] = max_wv*np.cos(wd_rad)
df['max Wy'] = ma... | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
The distribution of wind vectors is much simpler for the model to correctly interpret. | plt.hist2d(df['Wx'], df['Wy'], bins=(50, 50), vmax=400)
plt.colorbar()
plt.xlabel('Wind X [m/s]')
plt.ylabel('Wind Y [m/s]')
ax = plt.gca()
ax.axis('tight') | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
Time Similarly the `Date Time` column is very useful, but not in this string form. Start by converting it to seconds: | timestamp_s = date_time.map(datetime.datetime.timestamp) | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
Similar to the wind direction the time in seconds is not a useful model input. Being weather data it has clear daily and yearly periodicity. There are many ways you could deal with periodicity.A simple approach to convert it to a usable signal is to use `sin` and `cos` to convert the time to clear "Time of day" and "Ti... | day = 24*60*60
year = (365.2425)*day
df['Day sin'] = np.sin(timestamp_s * (2 * np.pi / day))
df['Day cos'] = np.cos(timestamp_s * (2 * np.pi / day))
df['Year sin'] = np.sin(timestamp_s * (2 * np.pi / year))
df['Year cos'] = np.cos(timestamp_s * (2 * np.pi / year))
plt.plot(np.array(df['Day sin'])[:25])
plt.plot(np.arr... | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
This gives the model access to the most important frequency features. In this case you knew ahead of time which frequencies were important. If you didn't know, you can determine which frequencies are important using an `fft`. To check our assumptions, here is the `tf.signal.rfft` of the temperature over time. Note the ... | fft = tf.signal.rfft(df['T (degC)'])
f_per_dataset = np.arange(0, len(fft))
n_samples_h = len(df['T (degC)'])
hours_per_year = 24*365.2524
years_per_dataset = n_samples_h/(hours_per_year)
f_per_year = f_per_dataset/years_per_dataset
plt.step(f_per_year, np.abs(fft))
plt.xscale('log')
plt.ylim(0, 400000)
plt.xlim([0.1... | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
Split the data We'll use a `(70%, 20%, 10%)` split for the training, validation, and test sets. Note the data is **not** being randomly shuffled before splitting. This is for two reasons.1. It ensures that chopping the data into windows of consecutive samples is still possible.2. It ensures that the validation/test re... | column_indices = {name: i for i, name in enumerate(df.columns)}
n = len(df)
train_df = df[0:int(n*0.7)]
val_df = df[int(n*0.7):int(n*0.9)]
test_df = df[int(n*0.9):]
num_features = df.shape[1] | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
Normalize the dataIt is important to scale features before training a neural network. Normalization is a common way of doing this scaling. Subtract the mean and divide by the standard deviation of each feature. The mean and standard deviation should only be computed using the training data so that the models have no a... | train_mean = train_df.mean()
train_std = train_df.std()
train_df = (train_df - train_mean) / train_std
val_df = (val_df - train_mean) / train_std
test_df = (test_df - train_mean) / train_std | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
Now peek at the distribution of the features. Some features do have long tails, but there are no obvious errors like the `-9999` wind velocity value. | df_std = (df - train_mean) / train_std
df_std = df_std.melt(var_name='Column', value_name='Normalized')
plt.figure(figsize=(12, 6))
ax = sns.violinplot(x='Column', y='Normalized', data=df_std)
_ = ax.set_xticklabels(df.keys(), rotation=90) | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
Data windowingThe models in this tutorial will make a set of predictions based on a window of consecutive samples from the data. The main features of the input windows are:* The width (number of time steps) of the input and label windows* The time offset between them.* Which features are used as inputs, labels, or bot... | class WindowGenerator():
def __init__(self, input_width, label_width, shift,
train_df=train_df, val_df=val_df, test_df=test_df,
label_columns=None):
# Store the raw data.
self.train_df = train_df
self.val_df = val_df
self.test_df = test_df
# Work out the label column... | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
Here is code to create the 2 windows shown in the diagrams at the start of this section: | w1 = WindowGenerator(input_width=24, label_width=1, shift=24,
label_columns=['T (degC)'])
w1
w2 = WindowGenerator(input_width=6, label_width=1, shift=1,
label_columns=['T (degC)'])
w2 | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
2. SplitGiven a list consecutive inputs, the `split_window` method will convert them to a window of inputs and a window of labels.The example `w2`, above, will be split like this:This diagram doesn'... | def split_window(self, features):
inputs = features[:, self.input_slice, :]
labels = features[:, self.labels_slice, :]
if self.label_columns is not None:
labels = tf.stack(
[labels[:, :, self.column_indices[name]] for name in self.label_columns],
axis=-1)
# Slicing doesn't preserve static s... | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
Try it out: | # Stack three slices, the length of the total window:
example_window = tf.stack([np.array(train_df[:w2.total_window_size]),
np.array(train_df[100:100+w2.total_window_size]),
np.array(train_df[200:200+w2.total_window_size])])
example_inputs, example_labels = w2.spl... | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
Typically data in TensorFlow is packed into arrays where the outermost index is across examples (the "batch" dimension). The middle indices are the "time" or "space" (width, height) dimension(s). The innermost indices are the features.The code above took a batch of 3, 7-timestep windows, with 19 features at each time s... | w2.example = example_inputs, example_labels
def plot(self, model=None, plot_col='T (degC)', max_subplots=3):
inputs, labels = self.example
plt.figure(figsize=(12, 8))
plot_col_index = self.column_indices[plot_col]
max_n = min(max_subplots, len(inputs))
for n in range(max_n):
plt.subplot(3, 1, n+1)
plt... | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
This plot aligns inputs, labels, and (later) predictions based on the time that the item refers to: | w2.plot() | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
You can plot the other columns, but the example window `w2` configuration only has labels for the `T (degC)` column. | w2.plot(plot_col='p (mbar)') | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
4. Create `tf.data.Dataset`s Finally this `make_dataset` method will take a time series `DataFrame` and convert it to a `tf.data.Dataset` of `(input_window, label_window)` pairs using the `preprocessing.timeseries_dataset_from_array` function. | def make_dataset(self, data):
data = np.array(data, dtype=np.float32)
ds = tf.keras.preprocessing.timeseries_dataset_from_array(
data=data,
targets=None,
sequence_length=self.total_window_size,
sequence_stride=1,
shuffle=True,
batch_size=32,)
ds = ds.map(self.split_window)
... | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
The `WindowGenerator` object holds training, validation and test data. Add properties for accessing them as `tf.data.Datasets` using the above `make_dataset` method. Also add a standard example batch for easy access and plotting: | @property
def train(self):
return self.make_dataset(self.train_df)
@property
def val(self):
return self.make_dataset(self.val_df)
@property
def test(self):
return self.make_dataset(self.test_df)
@property
def example(self):
"""Get and cache an example batch of `inputs, labels` for plotting."""
result = get... | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
Now the `WindowGenerator` object gives you access to the `tf.data.Dataset` objects, so you can easily iterate over the data.The `Dataset.element_spec` property tells you the structure, `dtypes` and shapes of the dataset elements. | # Each element is an (inputs, label) pair
w2.train.element_spec | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
Iterating over a `Dataset` yields concrete batches: | for example_inputs, example_labels in w2.train.take(1):
print(f'Inputs shape (batch, time, features): {example_inputs.shape}')
print(f'Labels shape (batch, time, features): {example_labels.shape}') | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
Single step modelsThe simplest model you can build on this sort of data is one that predicts a single feature's value, 1 timestep (1h) in the future based only on the current conditions.So start by building models to predict the `T (degC)` value 1h into the future.'])
single_step_window | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
The `window` object creates `tf.data.Datasets` from the training, validation, and test sets, allowing you to easily iterate over batches of data. | for example_inputs, example_labels in single_step_window.train.take(1):
print(f'Inputs shape (batch, time, features): {example_inputs.shape}')
print(f'Labels shape (batch, time, features): {example_labels.shape}') | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
BaselineBefore building a trainable model it would be good to have a performance baseline as a point for comparison with the later more complicated models.This first task is to predict temperature 1h in the future given the current value of all features. The current values include the current temperature. So start wit... | class Baseline(tf.keras.Model):
def __init__(self, label_index=None):
super().__init__()
self.label_index = label_index
def call(self, inputs):
if self.label_index is None:
return inputs
result = inputs[:, :, self.label_index]
return result[:, :, tf.newaxis] | _____no_output_____ | Apache-2.0 | site/en/tutorials/structured_data/time_series.ipynb | jcjveraa/docs-2 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.