import pandas as pd
import numpy as np
import warnings
import matplotlib.pyplot as plt
from statsmodels.tsa.arima.model import ARIMA

 %matplotlib inline
warnings.filterwarnings("ignore")

df = pd.read_excel("Dados.xlsx")

 def mape(actual, pred):
   return np.mean(np.abs((actual - pred) / actual)) * 100

df = df[["Data", "Vendas"]]
df.columns = ["Data", "Vendas"]
df["date"] = pd.to_datetime(df.Data)
df.set_index(df.Data, inplace = True)
df.sort_index(ascending = True, inplace = True)
df.drop("Data", axis = 1, inplace = True)
df.head()

	Vendas	date
Data
2022-12-06	870	2022-12-06
2022-12-07	868	2022-12-07
2022-12-08	1189	2022-12-08
2022-12-09	742	2022-12-09
2022-12-10	317	2022-12-10

plt.figure(figsize = (15, 10))
ax = plt.subplot(111)
ax.plot(df.index, df.Vendas)
ax.spines["top"].set_visible(False)
ax.spines["right"].set_visible(False)
ax.tick_params(labelsize = 16)
plt.show()

gr_dt = df.groupby(df.index.day).sum()
plt.figure(figsize = (15, 10))
ax = plt.subplot(111)
ax.plot(gr_dt.index, gr_dt.Vendas, color = "black", marker = "o")
ax.spines["top"].set_visible(False)
ax.spines["right"].set_visible(False)
ax.tick_params(labelsize = 16)
plt.show()

"""params = []
for x in range(0, 11):
    for y in range(0, 11):
        for z in range(0, 11):
            params.append((x, y, z))
best_param = None
best_acc = float("inf")
for param in params:
    try:
        model = ARIMA(df.Vendas, order = param).fit()
        acc = mape(df.Vendas, model.predict(typ = "levels"))
        if acc < best_acc:
            best_acc = acc
            best_param = param
        print("Order: ", param, "MAPE:", acc)
    except: 
        pass
print("Melhor order: ", best_param, " MAPE:", best_acc)"""

model = ARIMA(df.Vendas, order = (7,1,6)).fit()
acc = mape(df.Vendas, model.predict(typ = "levels"))

fig, ax = plt.subplots(figsize = (15, 10))
actual = ax.plot(df.Vendas, color = "deepskyblue")
pred = ax.plot(model.predict(typ = "levels"), color = "orangered")
ax.spines["top"].set_visible(False)
ax.spines["right"].set_visible(False)
ax.tick_params(labelsize = 16)
 
plt.show()

pred_steps = 5
pred_data = model.forecast(steps = pred_steps)


plt.figure(figsize = (15, 10))
ax = plt.subplot(111)
ax.plot(pred_data, color = "black", marker = "o")
ax.spines["top"].set_visible(False)
ax.spines["right"].set_visible(False)
ax.tick_params(labelsize = 16)


plt.show()