Disaster-Tweet-Analysis

This repo contains an approach I implemented for the Disaster Tweets competition on Kaggle.

Data

Each sample in the train and test set has the following information:

• The text of a tweet
• A keyword from that tweet (although this may be blank!)
• The location the tweet was sent from (may also be blank)

Files

• train.csv - the training set
• test.csv - the test set
• sample_submission.csv - a sample submission file in the correct format

Columns

• id - a unique identifier for each tweet
• text - the text of the tweet
• location - the location the tweet was sent from (may be blank)
• keyword - a particular keyword from the tweet (may be blank)
• target - in train.csv only, this denotes whether a tweet is about a real disaster (1) or not (0)

Necessary Libraires or packages are:-

• Pandas
• Numpy
• Regular expression (re)
• NLTK
• matplotlib
• tensorflow
• GloVe

My approach

• Clean text data present in both test and train dataset
• tokenizing the text
• removing the stopwords (on the basis of nltk.corpus.stopwords)
• lemmatizing the text entries
• converting list of strings into joint strings
• Importing the GloVe embeddings

Next, I have used TextVectorization layer. And I also create an embedding matrix. In this case, the embedding matrix is N-by-300 matrix, where N is the total count of distinct words in the input dataframe. If the word is not found in the embeddings_index, then the corresponding column is all-zeroes.

At last, I have initialized the keras model for training purpose, the architecture of the model is Embedding layer, Dropout, LSTM, Conv1D, GlobalMaxPooling1D, Dense and Output.