app.py

import streamlit as st
import pandas as pd 
import numpy as np 
import matplotlib.pyplot as plt 
from statsmodels.tsa.stattools import adfuller
from statsmodels.tsa.api import ExponentialSmoothing, SimpleExpSmoothing, Holt
from sklearn.metrics import mean_squared_error, mean_absolute_error
from math import sqrt

data= pd.read_csv('sample_dataset_timeseries_noarea.csv')
#menggunakan dictionary untuk menyimpan tabel dari setiap product
def table_generator(df, listname):
    name={}
    n= len(listname)
    for i in range(n):
       name[i]=df[df['product_item']==listname[i]]
    return name
# menghitung berapa banyak kejadian pembelian setiap produk
value_counts= data['product_item'].value_counts()

#membuat tabel baru dengan hanya produk yang dibeli di seluruh minggu
tabel_baru = data.loc[data['product_item'].isin(value_counts[value_counts == 67].index)]

# Kelompokkan data berdasarkan product_item dan hitung total quantity
grouped_data = tabel_baru.groupby('product_item')['quantity'].sum()

# Urutkan data secara menurun berdasarkan total quantity
sorted_data = grouped_data.sort_values(ascending=False)

# Ambil product_item dengan total quantity terbanyak
top_product_item = sorted_data.index[0]

sorted_data2= sorted_data.head(10)

product_item_names = sorted_data2.index.tolist()
new_data = data[data['product_item'].isin(product_item_names)]
name= table_generator(data, product_item_names)
week= new_data.groupby('week_number')['quantity'].sum()
# Ubah series menjadi dataframe
df = week.to_frame().reset_index()

# Ganti nama kolom dari indeks menjadi 'Week Number'
week = df.rename(columns={'index': 'Week Number'})
#splitting train and test
train=week[0:53] 
test=week[53:]
y_hat_avg = test.copy()
fit1 = ExponentialSmoothing(np.asarray(train['quantity']), seasonal_periods=13, trend='multiplicative', seasonal='multiplicative').fit()
y_hat_avg['Holt_Winter'] = fit1.forecast(len(test))

# Create the plot
fig, ax = plt.subplots(figsize=(20, 8))
ax.plot(train['quantity'], label='Train')
ax.plot(test['quantity'], label='Test')
ax.plot(y_hat_avg['Holt_Winter'], label='Holt_Winter')
ax.legend(loc='best')

# Show the plot in Streamlit
st.pyplot(fig)

def forecasting(x):
    predict= fit1.forecast(len(test)+x)
    predict= predict[-x:]
    rounded_arr = []
    for num in predict:
        rounded_num = round(num)
        rounded_arr.append(rounded_num)
    return rounded_arr
number = st.number_input("Enter a number between 1 and 10 to predict upcoming weeks", min_value=1, max_value=10)
num=forecasting(number)
st.write(num)
num1 = st.number_input("forecast a product", min_value=0, max_value=9)
def holtz_generator(df):
    df=df.drop(columns=['week_start_date','week_end_date','product_item'])
    n=0.8*len(df)
    n=round(n)
    #splitting train and test
    train=df.iloc[0:n] 
    test=df.iloc[n:]

    y_hat_avg = test
    fit1 = ExponentialSmoothing(np.asarray(train['quantity']) ,seasonal_periods=8 ,trend='multiplicative', seasonal='multiplicative').fit()
    y_hat_avg['Holt_Winter'] = fit1.forecast(len(test))
    # Create the plot
    fig, ax = plt.subplots(figsize=(20, 8))
    ax.plot(train['quantity'], label='Train')
    ax.plot(test['quantity'], label='Test')
    ax.plot(y_hat_avg['Holt_Winter'], label='Holt_Winter')
    ax.legend(loc='best')
    # Show the plot in Streamlit
    st.pyplot(fig)

holtz_generator(name[num1])