🔥Lit or Arson? Disaster Tweet Classification Part Two: Starting Feature Engineering & Selection

Recap of Part 1

In Part 1 we:

• Looked at the Kaggle problem
• Imported the data using Pandas
• Looked at high-level statistics of our dataset
• Checked for class imbalances
• Took a close look at the Keyword feature

Plan for Part 2

Now we’ll:

• Look at the Location feature
• Tweet character and token length
• Use Mann Whitney U test for feature selection

Part 1: Data Exploration

Part 2: Starting Feature Engineering & Selection — You’re here!

Location exploration

We’ll next take a look at the Location feature in a very similar fashion to what we did earlier for Keyword.

# Value counts
train_df['location'].value_counts(dropna=False)