import matplotlib
matplotlib.use("Qt5Agg")
import numpy as np
import matplotlib.pyplot as plt
import pandas as pd
from exploratoryUtils import basic_info_df
dataset_store = pd.read_csv("./data/store.csv", encoding="utf-8")
dataset_train = pd.read_csv("./data/train.csv", encoding="utf-8")
dataset_test = pd.read_csv("./data/test.csv", encoding="utf-8")
basic_info_df(dataset_store)
basic_info_df(dataset_train)
basic_info_df(dataset_test)
# Convert the column Date from object into datetime
dataset_train["Date"] = pd.to_datetime(dataset_train["Date"], format="%Y-%m-%d")
plt.plot_date(dataset_train.loc[dataset_train.Store == 1, "Date"], dataset_train.loc[dataset_train.Store == 1, "Sales"])
plt.title("Sales for Store 1 per date")
plt.savefig("./image/sales_store_1.png")
plt.show()
temp = dataset_train.groupby("Date")["Sales"]
dates = np.unique(dataset_train.loc[:, "Date"].values)
plt.plot_date(dates, temp.apply(lambda x: x.mean()), "ro", label="Mean")
import matplotlib
matplotlib.use('Qt5Agg')
import numpy as np
import matplotlib.pyplot as plt
import pandas as pd
from exploratoryUtils import basic_info_df
dataset_store = pd.read_csv("./data/store.csv", encoding="utf-8")
dataset_train = pd.read_csv("./data/train.csv", encoding="utf-8")
dataset_test = pd.read_csv("./data/test.csv", encoding="utf-8")
basic_info_df(dataset_store)
basic_info_df(dataset_train)
basic_info_df(dataset_test)
# Convert the column Date from object into datetime
dataset_train['Date'] = pd.to_datetime(dataset_train['Date'],
format='%Y-%m-%d')
plt.plot_date(dataset_train.loc[dataset_train.Store == 1, 'Date'],
dataset_train.loc[dataset_train.Store == 1, 'Sales'])
plt.title("Sales for Store 1 per date")
plt.savefig("./image/sales_store_1.png")
plt.show()
temp = dataset_train.groupby('Date')['Sales']
dates = np.unique(dataset_train.loc[:, 'Date'].values)
plt.plot_date(dates, temp.apply(lambda x: x.mean()), 'ro', label="Mean")
import matplotlib
matplotlib.use('Qt5Agg')
import matplotlib.pyplot as plt
import pandas as pd
from exploratoryUtils import basic_info_df
dataset_gift = pd.read_csv("./data/gifts.csv")
basic_info_df(dataset_gift)
# s for the size of the points / alpha for the transparency
dataset_gift.plot.scatter('Longitude',
'Latitude',
s=1,
color="yellow",
alpha=0.5)
plt.title("Gift location (%d gifts)" % len(dataset_gift))
plt.show()
plt.savefig("./image/gift_location.png")
