CNN-RNN-NLP

(1dCNN, 1dRNN) + NLP Project OverView

• Built several Machine Learning models (NLP) and Neural Network models (CNN, RNN) to predict types of meals do people eat base on foods name
• Built Feature Engineering methods to give data new feature.

Code and Resourced Used

• Python 3.8
• Packages: pandas, numpy, seaborn, matplotlib, sklearn, Xgboost, nltk, re, keras, tensorflow
• [Method] (https://zhuanlan.zhihu.com/p/62077601)
• [Embedding] (https://www.zhihu.com/question/45027109)

Data Preprocessing

1. Built function to remove all the words after ',' and '-' since most of they are food weight unit
2. one-hot encoding for y (label)
3. Combine food name and brand_name together, convert them to interger sequence

Fit and Predict models

1. NLP with XGboost classifier
2. CNN(1d)
3. RNN(1d)