Python libs I used most

search engine

NLTK: https://www.nltk.org/ (string processing: takes strings as input and returns strings or lists of strings as output)
spacy: https://spacy.io/ (object-oriented approach: returns document object whose words and sentences are objects themselves. Performance is better)
- spacy.displacy: https://spacy.io/usage/visualizers
gensim: https://radimrehurek.com/gensim/ (word2vec, doc2vec, LDA, distance metrics, text summarization, pivoted document length normalization)
flashtext: https://github.com/vi3k6i5/flashtext (fast, replace keywords in sentences or extract keywords from sentences)
bert-pytorch: https://github.com/codertimo/BERT-pytorch (train your own BERT model)
bert-as-service: https://github.com/hanxiao/bert-as-service (using BERT model as a sentence encoding service, i.e. mapping a variable-length sentence to a fixed-length vector)
- TensorFlow >= 1.10 and <=1.15 (don't install TensorFlow 2.0)
- /home/wlz/Utils/bert-serving-server --> server dir

pandas: https://pandas.pydata.org/
scipy: https://www.scipy.org/
numpy: https://numpy.org/
scikit-learn: https://scikit-learn.org/stable/
- categorical-encoding: https://github.com/scikit-learn-contrib/categorical-encoding (different encoding methods)
- imbalanced-learn: https://github.com/scikit-learn-contrib/imbalanced-learn#imbalanced-learn (under-sampling, over-sampling, combining over-/under-sampling, create ensemble balanced sets)
statsmodels: https://www.statsmodels.org/stable/index.html (cross-sectional models, time-series models)
prefixspan: https://github.com/chuanconggao/PrefixSpan-py (PrefixSpan: frequent sequential pattern mining, BIDE: frequent closed sequential pattern mining (in closed.py), FEAT: frequent generator sequential pattern mining algorithm (in generator.py))
modle explaination:
- lime: https://github.com/marcotcr/lime (explaining predictions for classifiers)
- SHAP: https://github.com/slundberg/shap (SHapley Additive exPlanations - a game theoretic approach to explain the output of any machine learning model)

sympy: https://www.sympy.org/en/index.html (symbolic mathematics, computer algebra system)

pytorch: https://pytorch.org/get-started/locally/
tensorflow: https://www.tensorflow.org/install/
- tf.keras: https://www.tensorflow.org/guide/keras/overview
  Keras users who use multi-backend Keras with the TensorFlow backend switch to tf.keras in TensorFlow 2.0. tf.keras is better maintained and has better integration with TensorFlow features (eager execution, distribution support and other)
```
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers
```
ann-visualizer: https://github.com/Prodicode/ann-visualizer (create a presentable graph of the neural network you are building)
- use tf.keras (did not install Keras)
- conda install python-graphviz instead of pip install graphviz
- modify modify the C:\conda3\Lib\site-packages\ann_visualizer\visualize.py file (see the example)

matplotlib: https://matplotlib.org/
seaborn: https://seaborn.pydata.org/ (based on matplotlib, provides a high-level interface)
bokeh: https://docs.bokeh.org/en/latest/index.html (interactive visualization)
plotly: https://plot.ly/python/ (basic charts, scientific charts, financial charts, maps, 3D charts, subplots, Jupyter Widgets Interaction, transforms, custom controls, animations)