cosine_species.py

import pandas as pd
import matplotlib.pyplot as plt
from sklearn.metrics.pairwise import cosine_similarity

import utils


def run():
    # List of species we want a map for
    SPECIES = utils.get_analysis_species()

    # Load bird sightings
    df = utils.load_ebird_data(filter=True, region='GE-MM')

    # Get a list of checklists and species
    checklists = list(df['SAMPLING EVENT IDENTIFIER'].unique())
    species = list(df['SCIENTIFIC NAME'].unique())

    # Build these into a matrix
    df_check = pd.DataFrame(index=checklists, columns=species)
    df_check = df_check.fillna(0)

    # Fill in the matrix cells if species appears in checklist
    for index, row in df.iterrows():
        spec = row['SCIENTIFIC NAME']
        chck = row['SAMPLING EVENT IDENTIFIER']
        df_check.loc[chck, spec] = 1

    # Create figure with two subplots
    fig, ax = plt.subplots(1, 2, figsize=(20,12))

    for index, bird in enumerate(SPECIES):
        col_bird = df_check[bird]
        col_other = df_check.copy().drop(bird, axis=1)
        similarities = []

        # Find similarities
        for col in col_other:
            similarities.append((col,
                                cosine_similarity(col_bird.values.reshape(1,-1), col_other[col].values.reshape(1,-1))[0][0]))
        
        # Sort by largest to smallest
        similarities.sort(key=lambda x: -x[1])

        # Get top 10
        similarities = similarities[:20]
        df_sim = pd.DataFrame(similarities, columns=['Scientific Name', 'Similarity'])

        # Find common name
        df_sim['Common Name'] = df_sim['Scientific Name'].apply(lambda x: df.loc[df['SCIENTIFIC NAME'] == x, 'COMMON NAME'].max())
        
        # Invert DataFrame
        df_sim = df_sim.iloc[::-1]

        # Plot bar chart
        df_sim.plot(kind='barh', ax=ax[index], y='Similarity', x='Common Name')
        ax[index].set_title(f'{bird} Similars')

    # Save image of similar birds
    fig.savefig(f'./fig/cosine_similars.png', bbox_inches='tight')


if __name__ == '__main__':
    run()