Script reads laji.fi data file, and a heatmap of weekly species distribution.

weekly_species_heatmaps.py

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

from sklearn.preprocessing import MinMaxScaler

rarity_threshold = 100 

# Read datafile that has number of observations for each species
df = pd.read_csv('species_rarity.csv', sep='\t')
species_rarity = pd.Series(df['count'].values, index=df['sciname']).to_dict()


def remove_long_gatherings(row):
    if row['Gathering.Conversions.DayOfYearEnd'] > row['Gathering.Conversions.DayOfYearBegin'] + 3:
        return False
    return True


def remove_non_species(row):
    species = row['Taxon.ScientificName']
    if not isinstance(species, str):
        return False
    if " " in species:
        return True
    return False


def remove_rarities(row):
    if row['Taxon.ScientificName'] not in species_rarity:
        return False
    if species_rarity[row['Taxon.ScientificName']] < rarity_threshold:
        return False
    else:
        return True


def remove_commons(row):
    if row['Taxon.ScientificName'] not in species_rarity:
        return True
    if species_rarity[row['Taxon.ScientificName']] >= rarity_threshold:
        return False
    else:
        return True


# Read an observation datafile 
data = pd.read_csv('luteet-suomi-2000.csv', sep=';')
size_x = 20
size_y = 130
fontsize = 8

# Filtering the data
data = data[data.apply(remove_rarities, axis=1)]
data = data[data.apply(remove_long_gatherings, axis=1)]
data = data[data.apply(remove_non_species, axis=1)]

medians = data.groupby('Taxon.ScientificName')['Gathering.Conversions.DayOfYearBegin'].median().sort_values()

# convert 'day_of_year' to week_of_year
data['week_of_year'] = data['Gathering.Conversions.DayOfYearBegin'] // 7

# create a new dataframe where the rows are species, the columns are weeks of the year, 
# and the values are the count of observations
heatmap_data = data.groupby(['Taxon.ScientificName', 'week_of_year']).size().unstack(fill_value=0)

# Align the 'heatmap_data' index with 'medians' index.
heatmap_data = heatmap_data.reindex(medians.index)

# plot the heatmap
plt.figure(figsize=(size_x, size_y))
ax = sns.heatmap(heatmap_data, cmap='vlag')
ax.tick_params(axis='y', labelsize=fontsize)

# set labels
plt.xlabel("Week of Year")
plt.ylabel("Species")
plt.title("Observations of Species over Time")

# save the plot as an image file
plt.savefig('heatmap.png', dpi=300)