Demand-ForeCasting

import pandas as pd import streamlit as st import plotly.express as px from plotly import graph_objs as go st.title("Demand Trend Analysis")

df = pd.read_csv("data/cleaned_data.csv",parse_dates=['Order Date'],index_col='Order Date') df_train = df.index< '2018-01-01'

df_test = df.index>= '2018-01-01' df_train = df[df_train] df_test = df[df_test] time_pred = ["Past","Future"]

#display the years of data as a slider 2015-2017 for past and 2018 for future

k = st.sidebar.selectbox("Time",time_pred) if k == "Past": n_years = st.sidebar.slider("Years of data", 2015, 2016, 2017)

periods = 12*n_years

else: n_years = st.sidebar.slider("Years of data", 2018,2019) periods = 12

@st.cache_data def load_data(): data = df.copy()

return data

data_load_state = st.text("Loading data...") data = load_data() data_load_state.text("Loading data...done!")

st.subheader("Raw data") st.write(data.head())

def plot_raw_data_year(input:str):

if input == "Past":
    
    df_yearly= df_train.groupby(pd.Grouper(freq='Y'))['Sales'].sum()
    df_yearly = pd.DataFrame(df_yearly)
else:
    df_yearly = df_test.groupby(pd.Grouper(freq='Y'))['Sales'].sum()
    df_yearly = pd.DataFrame(df_yearly)
    
fig = go.Figure()
fig.add_trace(go.Bar(x=df_yearly.index, y=df_yearly.Sales,name='Yearly Sales' ,))
fig.update_layout(title_text='Yearly Sales',plot_bgcolor='white',xaxis_rangeslider_visible=True)
st.plotly_chart(fig)

plot_raw_data_year(k)

def plot_raw_data_month(input:str): if input == "Past": df_monthly= df_train.groupby(pd.Grouper(freq='M'))['Sales'].sum() df_monthly = pd.DataFrame(df_monthly) else: df_monthly = df_test.groupby(pd.Grouper(freq='M'))['Sales'].sum() df_monthly = pd.DataFrame(df_monthly)

fig = go.Figure()
fig.add_trace(go.Scatter(x=df_monthly.index, y=df_monthly.Sales,name='Monthly Sales' ))
fig.update_layout(title_text= 'Monthly Sales',plot_bgcolor='white',xaxis_rangeslider_visible=True)
st.plotly_chart(fig)

plot_raw_data_month(k)

def plot_raw_data_day(input:str): if input == "Past": df_daily= df_train.groupby(pd.Grouper(freq='D'))['Sales'].sum() df_daily = pd.DataFrame(df_daily) else: df_daily = df_test.groupby(pd.Grouper(freq='D'))['Sales'].sum() df_daily = pd.DataFrame(df_daily)

fig = go.Figure()
fig.add_trace(go.Scatter(x=df_daily.index, y=df_daily.Sales,name='Daily Sales' ))
fig.update_layout(title_text= 'Daily Sales',plot_bgcolor='white',xaxis_rangeslider_visible=True)
st.plotly_chart(fig)

plot_raw_data_day(k)

def plot_raw_yearly_sales_by_segment(input:str):

if input == "Past":
    df_yearly_segment = df_train.groupby([pd.Grouper(freq='Y'), 'Segment'])['Sales'].sum().reset_index()

  
    df_yearly_segment = pd.DataFrame(df_yearly_segment)
else:
    df_yearly_segment = df_test.groupby([pd.Grouper(freq='Y'), 'Segment'])['Sales'].sum().reset_index()

   
    df_yearly_segment = pd.DataFrame(df_yearly_segment)
color_scale = px.colors.sequential.Viridis

create a dictionary that maps each unique value in the Segment column to a color from the color scheme

color_map = {segment: color_scale[i % len(color_scale)] for i, segment in enumerate(df_yearly_segment['Segment'].unique())}

use the color_map dictionary to map the Segment values to colors

colors = df_yearly_segment['Segment'].map(color_map)

create the plot using plotly.graph_objects

fig = go.Figure(data=go.Bar(x=df_yearly_segment['Order Date'], y=df_yearly_segment['Sales'], marker={'color': colors},hovertext=df_yearly_segment['Segment']))
fig.update_layout(title_text='Yearly Sales by Segment', plot_bgcolor='white')

st.plotly_chart(fig)

plot_raw_yearly_sales_by_segment(k) def plot_raw_yearly_sales_by_region(input:str):

if input == "Past":
    df_yearly_segment = df_train.groupby([pd.Grouper(freq='Y'), 'Region'])['Sales'].sum().reset_index()

  
    df_yearly_segment = pd.DataFrame(df_yearly_segment)
else:
    df_yearly_segment = df_test.groupby([pd.Grouper(freq='Y'), 'Region'])['Sales'].sum().reset_index()

   
    df_yearly_segment = pd.DataFrame(df_yearly_segment)
color_scale = px.colors.sequential.Viridis

create a dictionary that maps each unique value in the Segment column to a color from the color scheme

color_map = {segment: color_scale[i % len(color_scale)] for i, segment in enumerate(df_yearly_segment['Region'].unique())}

use the color_map dictionary to map the Segment values to colors

colors = df_yearly_segment['Region'].map(color_map)

create the plot using plotly.graph_objects

fig = go.Figure(data=go.Bar(x=df_yearly_segment['Order Date'], y=df_yearly_segment['Sales'], marker={'color': colors},hovertext=df_yearly_segment['Region']))
fig.update_layout(title_text='Yearly Sales by  Region', plot_bgcolor='white')
st.plotly_chart(fig)

plot_raw_yearly_sales_by_region(k)

def plot_raw_yearly_sales_by_Category(input:str):

if input == "Past":
    df_yearly_segment = df_train.groupby([pd.Grouper(freq='Y'), 'Category'])['Sales'].sum().reset_index()

  
    
else:
    df_yearly_segment = df_test.groupby([pd.Grouper(freq='Y'), 'Category'])['Sales'].sum().reset_index()

   
df_yearly_segment = pd.DataFrame(df_yearly_segment)
color_scale = px.colors.sequential.Viridis

create a dictionary that maps each unique value in the Segment column to a color from the color scheme

color_map = {segment: color_scale[i % len(color_scale)] for i, segment in enumerate(df_yearly_segment['Category'].unique())}

use the color_map dictionary to map the Segment values to colors

colors = df_yearly_segment['Category'].map(color_map)

create the plot using plotly.graph_objects

fig = go.Figure(data=go.Bar(x=df_yearly_segment['Order Date'], y=df_yearly_segment['Sales'], marker={'color': colors},hovertext=df_yearly_segment['Category']))
fig.update_layout(title_text='Yearly Sales by  Category', plot_bgcolor='white')
st.plotly_chart(fig)

plot_raw_yearly_sales_by_Category(k)

def plot_raw_yearly_sales_by_State(input:str, number:int):

if input == "Past":
    df_yearly_state = df_train.groupby([pd.Grouper(freq='Y'), 'State'])['Sales'].sum().reset_index()
else:
    df_yearly_state = df_test.groupby([pd.Grouper(freq='Y'), 'State'])['Sales'].sum().reset_index()
    
df_yearly_state = pd.DataFrame(df_yearly_state)
color_scale = px.colors.sequential.Viridis
topN_states = df_yearly_state.groupby('State').sum().sort_values('Sales', ascending=False).head(number).index.tolist()
top_states_df = df_yearly_state[df_yearly_state['State'].isin(topN_states)]

# create a dictionary that maps each unique value in the State column to a color from the color scheme
color_map = {state: color_scale[i % len(color_scale)] for i, state in enumerate(top_states_df['State'].unique())}

# use the color_map dictionary to map the State values to colors
colors = top_states_df['State'].map(color_map)

# create the plot using plotly.graph_objects
fig = go.Figure(data=go.Bar(x=top_states_df['Order Date'], y=top_states_df['Sales'], marker={'color': colors},hovertext=top_states_df['State']))
fig.update_layout(title_text=f'Top {number} states with highest sales', plot_bgcolor='white')
st.plotly_chart(fig)

initialize Streamlit slider for selecting number of subcategories to display

number_st = st.slider('Select the number of States', 1, 10, 3)

plot_raw_yearly_sales_by_State(k,number_st)

def plot_raw_yearly_sales_by_Sub_Cat(input:str, number:int):

if input == "Past":
    df_yearly_state = df_train.groupby([pd.Grouper(freq='Y'), 'Sub-Category'])['Sales'].sum().reset_index()
else:
    df_yearly_state = df_test.groupby([pd.Grouper(freq='Y'), 'Sub-Category'])['Sales'].sum().reset_index()
    
df_yearly_state = pd.DataFrame(df_yearly_state)
color_scale = px.colors.sequential.Viridis
topN_states = df_yearly_state.groupby('Sub-Category').sum().sort_values('Sales', ascending=False).head(number).index.tolist()
top_states_df = df_yearly_state[df_yearly_state['Sub-Category'].isin(topN_states)]

# create a dictionary that maps each unique value in the State column to a color from the color scheme
color_map = {state: color_scale[i % len(color_scale)] for i, state in enumerate(top_states_df['Sub-Category'].unique())}

# use the color_map dictionary to map the State values to colors
colors = top_states_df['Sub-Category'].map(color_map)

# create the plot using plotly.graph_objects
fig = go.Figure(data=go.Bar(x=top_states_df['Order Date'], y=top_states_df['Sub-Category'], marker={'color': colors},hovertext=top_states_df['Sub-Category']))
fig.update_layout(title_text=f'Top {number} sub categories with highest sales', plot_bgcolor='white')
st.plotly_chart(fig)

initialize Streamlit slider for selecting number of subcategories to display

number_sub_cat = st.slider('Select the number of Sub-Category', 1, 10, 3)

plot_raw_yearly_sales_by_Sub_Cat(k,number_sub_cat)

def plot_raw_yearly_sales_by_Product(input:str,number:int):

if input == "Past":
    df_yearly_product = df_train.groupby([pd.Grouper(freq='Y'), 'Product Name'])['Sales'].sum().reset_index()
else:
    df_yearly_product = df_test.groupby([pd.Grouper(freq='Y'), 'Product Name'])['Sales'].sum().reset_index()
   
df_yearly_product = pd.DataFrame(df_yearly_product)
color_scale = px.colors.sequential.Viridis
topN_products = df_yearly_product.groupby('Product Name').sum().sort_values('Sales', ascending=False).head(number).index.tolist()
top_product_df = df_yearly_product[df_yearly_product['Product Name'].isin(topN_products)]

# create a dictionary that maps each unique value in the Product Name column to a color from the color scheme
color_map = {product: color_scale[i % len(color_scale)] for i, product in enumerate(top_product_df['Product Name'].unique())}

# use the color_map dictionary to map the Product Name values to colors
colors = top_product_df['Product Name'].map(color_map)

# create the plot using plotly.graph_objects
fig = go.Figure(data=go.Bar(x=top_product_df['Order Date'], y=top_product_df['Sales'], marker={'color': colors},hovertext=top_product_df['Product Name']))
fig.update_layout(title_text=f'Top {number} best-selling products', plot_bgcolor='white')
st.plotly_chart(fig)

initialize Streamlit slider for selecting number of products to display

number_p = st.slider('Select the number of products to display', 1, 10, 3) plot_raw_yearly_sales_by_Product(k,number_p)

def plot_raw_yearly_sales_by_City(input:str, number:int):

if input == "Past":
    df_yearly_state = df_train.groupby([pd.Grouper(freq='Y'), 'City'])['Sales'].sum().reset_index()
else:
    df_yearly_state = df_test.groupby([pd.Grouper(freq='Y'), 'City'])['Sales'].sum().reset_index()
    
df_yearly_state = pd.DataFrame(df_yearly_state)
color_scale = px.colors.sequential.Viridis
topN_states = df_yearly_state.groupby('City').sum().sort_values('Sales', ascending=False).head(number).index.tolist()
top_states_df = df_yearly_state[df_yearly_state['City'].isin(topN_states)]

# create a dictionary that maps each unique value in the State column to a color from the color scheme
color_map = {state: color_scale[i % len(color_scale)] for i, state in enumerate(top_states_df['City'].unique())}

# use the color_map dictionary to map the State values to colors
colors = top_states_df['City'].map(color_map)

# create the plot using plotly.graph_objects
fig = go.Figure(data=go.Bar(x=top_states_df['Order Date'], y=top_states_df['City'], marker={'color': colors},hovertext=top_states_df['City']))
fig.update_layout(title_text=f'Top {number} states with highest sales', plot_bgcolor='white')
st.plotly_chart(fig)

initialize Streamlit slider for selecting number of subcategories to display

number_city = st.slider('Select the number of Cities', 1, 10, 3)

plot_raw_yearly_sales_by_City(k,number_city)