optilearn 1.2.6

Use to train neural networks, A Package for optimize models, transfer or copy files from one directory to other, use for nlp short word treatment, choosing optimal data for ML models, use for Image Scraping , use in timeseries problem to split the data into train and test, Deal with emojis and emoticons in nlp, word tokenize, token, get the list of Punctuation marks and English Pronouns too, can be used to read text files

A package to increase the accuracy of ML models, training neural networks, converting text into tokens, transfer or copy files from one directory to other, gives you the best data for model training, works on text data also, short word treatment for NLP problems, can be used for Image Scraping also, use it to split the timeseries data into training and testing, deal with emojis and emoticons, word tokenizer, tokenize words, can be used to remove stop words too, Punctuations and get English Pronouns list too, use to read text files # Optilearn

Optilearn is a versatile package designed to improve the accuracy of machine learning models, handle neural network training, and offer various utilities for NLP, image scraping, and timeseries data manipulation.

Features:

• Neural Network Training: Use the optilearn.nn module to train neural networks.
• Optimal Data Combination: Use optilearn.optimal to find the best data combination for ML and DL models.
• Image Scraping: Use optilearn.images to download images from websites.
• Text and NLP Processing: Use optilearn.text and optilearn.words for text tokenization, short word treatment, and more NLP tasks.
• File Handling: Use optilearn.files to transfer and copy files between directories.
• Timeseries Data Handling: Use optilearn.sequential to split sequential data like timeseries into training and testing sets.

Installation

!pip install optilearn

1. Neural network training

from optilearn.nn import Sequential,Dense,Dropout,EarlyStopping,ReduceLearningRate

model2=Sequential() model2.add(Dense(300,input_dim=784,activation='relu')) model2.add(Dropout(0.3)) model2.add(Dense(100,activation='relu')) model2.add(Dropout(0.3)) model2.add(Dense(150,activation='relu')) model2.add(Dropout(0.3)) model2.add(Dense(100,activation='relu')) model2.add(Dropout(0.3)) model2.add(Dense(50,activation='relu')) model2.add(Dropout(0.1)) model2.add(Dense(10,activation='softmax'))

model2.compile(loss='categorical_crossentropy',optimizer='adam',learning_rate=0.01)

rl= ReduceLearningRate(monitor='val_accuracy',factor=0.1,patience=20,verbose=1,min_delta=0.001,cooldown=3,min_lr=0.0001) es = EarlyStopping(monitor = 'val_accuracy',min_delta=0.0001,mode='max',verbose=1,patience=30,restore_best_weights=True,baseline=0.99)

g1=model2.fit(x_train,y_train,validation_data=[x_test,y_test],epochs=500,verbose=1,batch_size=128,callbacks=[rl,es])

2. Get best data combination for ML and DL models

from optilearn.optimal import OptimalDataSelector

x_train,y_train,x_test,y_test = OptimalDataSelector(X,Y,combination=15000,train_size=0.8,active_checkpoint=True,bs_problem='reg',scaling='st_normal')

3. Dealing NLP Problems

from optilearn.text import EmoTextHandler,word_tokenizer,Punctuation,EnglishPronouns,read_txt,text_to_tokens from optilearn.word import short_word_treatment`

4. File Transfering from one to another folder

from optilearn.files import FilePathway

5. Image scrapping from websites

from optilearn.images import web_image_downloader

6. splitting timeseries based data into training and testing

from optilearn.sequential import timeseries_split

x_train,x_test,y_train,y_test=timeseries_split(df1,'08:00:00',n_previous_values=50,train_size=0.8) x_train.shape,x_test.shape,y_train.shape,y_test.shape

*** For more detailed documentation, visit our GitHub Page --> https://github.com/about-optilearn.