#DATA CLEANING CODE =.py

#DATA CLEANING CODE =

import pandas as pd
import dask.dataframe as dd

#This loads datasets using Dask
df = dd.read_csv("Trips_By_Distance.csv")
df_full = dd.read_csv("Trips_Full_Data.csv")

df = df.fillna(df.mean())
df_full = df_full.fillna(df_full.mean())

#Convert data types
df["Population Staying at Home"] = df["Population Staying at Home"].astype("int64")
df_full["Week"] = df_full["Week"].astype("int64")

df = df.drop_duplicates().compute()
df_full = df_full.drop_duplicates().compute()

print(df.head())
print(df_full.head())

print(df.isna().sum().compute())
print(df_full.isna().sum().compute())