import datetime as dt
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import seaborn as sbs
from matplotlib.colors import ListedColormap
from sklearn import neighbors, datasets
from sklearn.metrics import confusion_matrix, classification_report, mean_squared_error
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler

# https://www.kaggle.com/rtatman/did-it-rain-in-seattle-19482017/downloads/did-it-rain-in-seattle-19482017.zip/1
df = pd.read_csv('./seattleWeather_1948-2017.csv')
df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 25551 entries, 0 to 25550
Data columns (total 5 columns):
DATE    25551 non-null object
PRCP    25548 non-null float64
TMAX    25551 non-null int64
TMIN    25551 non-null int64
RAIN    25548 non-null object
dtypes: float64(1), int64(2), object(2)
memory usage: 998.2+ KB

data = pd.DataFrame()
data['temp_max'] = df.TMAX
data['temp_min'] = df.TMIN
data['date'] = pd.to_datetime(df.DATE)
data['precip'] = pd.to_numeric(df.PRCP)
data['above_1'] = data.precip > 1
data['above_avg'] = data.precip > 0.25
data['any_rain'] = data.precip > 0
data['prev_temp_max'] = [data.temp_max[0]] + list(data.temp_max[:-1])
data['prev_temp_min'] = [data.temp_min[0]] + list(data.temp_min[:-1])

data.plot(x='date', y='precip')

<matplotlib.axes._subplots.AxesSubplot at 0x112597630>

data.plot(x='date', y='temp_min')

<matplotlib.axes._subplots.AxesSubplot at 0x112d9ceb8>

data.describe()

	temp_max	temp_min	precip	prev_temp_max	prev_temp_min
count	25551.000000	25551.000000	25548.000000	25551.000000	25551.000000
mean	59.544206	44.514226	0.106222	59.544245	44.514461
std	12.772984	8.892836	0.239031	12.772956	8.892690
min	4.000000	0.000000	0.000000	4.000000	0.000000
25%	50.000000	38.000000	0.000000	50.000000	38.000000
50%	58.000000	45.000000	0.000000	58.000000	45.000000
75%	69.000000	52.000000	0.100000	69.000000	52.000000
max	103.000000	71.000000	5.020000	103.000000	71.000000

data.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 25551 entries, 0 to 25550
Data columns (total 9 columns):
temp_max         25551 non-null int64
temp_min         25551 non-null int64
date             25551 non-null datetime64[ns]
precip           25548 non-null float64
above_1          25551 non-null bool
above_avg        25551 non-null bool
any_rain         25551 non-null bool
prev_temp_max    25551 non-null int64
prev_temp_min    25551 non-null int64
dtypes: bool(3), datetime64[ns](1), float64(1), int64(4)
memory usage: 1.2 MB

sbs.pairplot(data.drop(columns=['precip']))

<seaborn.axisgrid.PairGrid at 0x1135e2eb8>

X = data.drop(columns=['above_1', 'above_avg', 'any_rain'])
y_above_1 = data['above_1']
y_above_avg = data['above_avg']
y_any_rain = data['any_rain']

X_train, X_test, y_above_1_train, y_above_1_test = train_test_split(X, y_above_1, test_size=0.3, random_state=123)
X_train, X_test, y_above_avg_train, y_above_avg_test = train_test_split(X, y_above_avg, test_size=0.3, random_state=123)
X_train, X_test, y_any_rain_train, y_any_rain_test = train_test_split(X, y_any_rain, test_size=0.3, random_state=123)