Superhero_NLP

"Work in Progress"

Superheros come to save the day! Source: Marjan Blan, unsplash.com

Intro

Trying to determine if a superhero is good or bad from text. A Natural Language Processing Model based on a dataset from Kaggle Kernels.

README Outline

• Introduction
• README Outline
• Repo Contents
• Libraries and Prerequisites
• Feature and Definitions
• Models
• Conclusions
• Future Work
• Built With, Contributors, Authors, Acknowledgments

Using 'text' not 'photographs' to determine who is who in the hero world! Source: Marjan Blan, Unsplash.com

Repo Contents

This repo contains the following:

• README.md - this is where you are now!
• Superhero_NLP_Notebook.ipynb - the Jupyter Notebook containing the finalized code for this project.
• LICENSE.md - the required license information.
• Data - superheroes_nlp_dataset.csv
• CONTRIBUTING.md
• Images - contains the fun images to this repo.
• gitignore

Libraries & Prerequisites

These are the libraries that I used in this project.

• import pandas as pd

• import numpy as np

• import matplotlib.pyplot as plt

• %matplotlib inline

• import seaborn as sns

• import string, re

• import nltk

• from nltk import FreqDist, word_tokenize

• from nltk.corpus import stopwords

• from sklearn.feature_extraction.text import TfidfVectorizer

• from sklearn.model_selection import train_test_split

• from sklearn.metrics import accuracy_score

• from sklearn.ensemble import RandomForestClassifier

• from sklearn.naive_bayes import MultinomialNB

Bar graph of Superheros ranked by total powers. Source: Superhero_NLP_Notebook.ipynb

Features

There are 81 columns in this dataset. Below is a sample of some of the ones I used:

• 0 name 1448 non-null object
• 1 real_name 1301 non-null object
• 2 full_name 956 non-null object
• 3 overall_score 1450 non-null object
• 4 history_text 1360 non-null object
• 5 powers_text 1086 non-null object
• 6 intelligence_score 1450 non-null int64
• 7 strength_score 1450 non-null int64
• 8 speed_score 1450 non-null int64
• 9 durability_score 1450 non-null int64
• 10 power_score 1450 non-null int64
• 11 combat_score 1450 non-null int64
• 12 superpowers 1450 non-null object
• 13 alter_egos 1450 non-null object
• 14 aliases 1450 non-null object
• 15 place_of_birth 788 non-null object
• 16 first_appearance 1247 non-null object
• 17 creator 1311 non-null object
• 18 alignment 1368 non-null object The majority of the rest of the features are floats, describing the superhero powers as a number.

Models

Naive Bayes and Random Forest

Conclusions

Future Work

There is so much more I want to do! I still have a lot to learn about text pre-processing - including how to put it all into a nice easy function.

Is this one of the good guys! Source: Yulia Matvienko, unsplash.com

Built With:

Jupyter Notebook Python 3.0 scikit.learn

Contributing

Please read CONTRIBUTING.md for details

Authors

Thomas Whipple

License

Please read LICENSE.md for details

Acknowledgments

Kaggle - This dataset comes from the following: https://www.kaggle.com/jonathanbesomi/superheroes-nlp-dataset