WALMART UNIT SALES PREDICTION — A Time Series ... https://m-mazumdar13.medium.com/walmart-unit-sales-predi...

1 of 24 11/8/2021, 2:32 PM
WALMART UNIT SALES PREDICTION — A Time Series ... https://m-mazumdar13.medium.com/walmart-unit-sales-predi...

2 of 24 11/8/2021, 2:32 PM
WALMART UNIT SALES PREDICTION — A Time Series ... https://m-mazumdar13.medium.com/walmart-unit-sales-predi...

3 of 24 11/8/2021, 2:32 PM
WALMART UNIT SALES PREDICTION — A Time Series ... https://m-mazumdar13.medium.com/walmart-unit-sales-predi...

4 of 24 11/8/2021, 2:32 PM
WALMART UNIT SALES PREDICTION — A Time Series ... https://m-mazumdar13.medium.com/walmart-unit-sales-predi...

5 of 24 11/8/2021, 2:32 PM
WALMART UNIT SALES PREDICTION — A Time Series ... https://m-mazumdar13.medium.com/walmart-unit-sales-predi...

6 of 24 11/8/2021, 2:32 PM
 1 sales_ca=sales_df[sales_df['state_id']=='CA'].loc[:,'d_1':].sum(axis=0)
2 sales_wi=sales_df[sales_df['state_id']=='WI'].loc[:,'d_1':].sum(axis=0)
3 sales_tx=sales_df[sales_df['state_id']=='TX'].loc[:,'d_1':].sum(axis=0)
4
5 state_sales_df=pd.DataFrame({'sales_ca':list(sales_ca),
6 'sales_wi':list(sales_wi),
7 'sales_tx':list(sales_tx),
8 'date':list(calendar_df['date']),
9 'weekday':calendar_df['weekday'],
10 'wday':calendar_df['wday'],
11 'month':calendar_df['month'],
12 'event_type_1':calendar_df['event_type_1'],
13 'event_type_2':calendar_df['event_type_2']})
14
15
16 fig=go.Figure()
17 fig.add_trace(go.Scatter(y=state_sales_df['sales_ca'],x=state_sales_df['date'
18 fig.add_trace(go.Scatter(y=state_sales_df['sales_wi'],x=state_sales_df['date'
19 fig.add_trace(go.Scatter(y=state_sales_df['sales_tx'],x=state_sales_df['date'
20 fig.update_layout(title='Aggregate sales per state',
WALMART UNIT SALES PREDICTION — A Time Series ... https://m-mazumdar13.medium.com/walmart-unit-sales-predi...

1 state_ca_trend=seasonal_decompose(state_sales_df['sales_ca'],freq=180).trend
2 state_wi_trend=seasonal_decompose(state_sales_df['sales_wi'],freq=180).trend
3 state_tx_trend=seasonal_decompose(state_sales_df['sales_tx'],freq=180).trend
4
5 fig=go.Figure()
6 fig.add_trace(go.Scatter(y=state_ca_trend,x=state_sales_df['date'],mode='lines'
7 fig.add_trace(go.Scatter(y=state_wi_trend,x=state_sales_df['date'],mode='lines'
8 fig.add_trace(go.Scatter(y=state_tx_trend,x=state_sales_df['date'],mode='lines'
9 fig.update_layout(title='Sales trend per state',
10 xaxis_title='year',
11 yaxis_title='sales')

8 of 24 11/8/2021, 2:32 PM
WALMART UNIT SALES PREDICTION — A Time Series ... https://m-mazumdar13.medium.com/walmart-unit-sales-predi...

1 from tqdm import tqdm

2 df=sales_df.iloc[:,6:]
3 x=set(np.array(df).flatten())
4 full_sales_array=np.array(df).flatten()
5 y=[]
6
7 for i in tqdm(x):
8 y.append(np.count_nonzero(df==i)/len(full_sales_array))
9
10 fig=go.Figure()
11 fig.add_trace(go.Bar(x=[np.count_nonzero(df==0)],y=['zero unit sales'],orientation
12 fig.add_trace(go.Bar(x=[np.count_nonzero(df)],y=['nonzero unit sales'],orientation
13 fig.update_layout(width=700,
14 height=300,
15 xaxis_title='Count')
16
17 fig1=px.bar(x=list(x),y=y,title='Probability density of unit sales figures')
18 fig1.update_layout(xaxis_title='unit sales figures',
19 yaxis_title='probability')
20

10 of 24 11/8/2021, 2:32 PM
WALMART UNIT SALES PREDICTION — A Time Series ... https://m-mazumdar13.medium.com/walmart-unit-sales-predi...

1 sales_states_weekly=state_sales_df.groupby(by=['weekday','wday'],
2 as_index=False)['sales_ca',
3 'sales_wi',
4 'sales_tx'].agg('mean'
5
6 fig=make_subplots(rows=1,cols=3)
7 fig.append_trace(go.Bar(x=sales_states_weekly['weekday'],y=sales_states_weekly
8 fig.append_trace(go.Bar(x=sales_states_weekly['weekday'],y=sales_states_weekly
9 fig.append_trace(go.Bar(x=sales_states_weekly['weekday'],y=sales_states_weekly
10 fig.update_layout(
11 title='Walmart weekly sales',
12 xaxis_title='weekday',
13 yaxis_title='Average sale')

11 of 24 11/8/2021, 2:32 PM
 1 sales_states_monthly=state_sales_df.groupby(by=['month'],
2 as_index=False
3 )['sales_ca',
4 'sales_wi',
5 'sales_tx'].agg('mean')
6
7 sales_states_monthly['month_name']=['Jan','Feb','Mar','Apr','May','June','July'
8 fig=make_subplots(rows=1,cols=3)
9 fig.append_trace(go.Line(x=sales_states_monthly['month_name'],y=sales_states_monthly
10 fig.append_trace(go.Line(x=sales_states_monthly['month_name'],y=sales_states_monthly
11 fig.append_trace(go.Line(x=sales_states_monthly['month_name'],y=sales_states_monthly
12 fig.update_layout(
13 title='Walmart monthly sales',
14 xaxis_title='month',
WALMART UNIT SALES PREDICTION — A Time Series ... https://m-mazumdar13.medium.com/walmart-unit-sales-predi...

1 walmart_sales=sales_df.loc[:,'d_1':].sum(axis=0)
2 plot_acf(walmart_sales,lags=60,title='Autocorrelation sales')
3 plt.ylabel('correlation')
4 plt.xlabel('time lag')
5 plt.show()

14 of 24 11/8/2021, 2:32 PM
WALMART UNIT SALES PREDICTION — A Time Series ... https://m-mazumdar13.medium.com/walmart-unit-sales-predi...

15 of 24 11/8/2021, 2:32 PM
WALMART UNIT SALES PREDICTION — A Time Series ... https://m-mazumdar13.medium.com/walmart-unit-sales-predi...

1 total_sales_hobbies=sales_df[sales_df['cat_id']=='HOBBIES'].loc[:,'d_1':].sum
2 total_sales_household=sales_df[sales_df['cat_id']=='HOUSEHOLD'].loc[:,'d_1':].
3 total_sales_foods=sales_df[sales_df['cat_id']=='FOODS'].loc[:,'d_1':].sum(axis
4
5 category_sales_df=pd.DataFrame({'sales_foods':list(total_sales_foods),
6 'sales_hobbies':list(total_sales_hobbies),
7 'sales_household':list(total_sales_household),
8 'date':list(calendar_df['date']),
9 'event_type_1':calendar_df['event_type_1'],
10 'event_type_2':calendar_df['event_type_2']},
11 )
12 no_event_sales_category=category_sales_df.iloc[:,:3][category_sales_df['event_type_1'
13 cat_sales_on_events=category_sales_df.groupby(['event_type_1'],as_index=False
14 'sales_hobbies',
15 'sales_household'].
16
17 cat_sales_on_events['sales_foods_diff']=cat_sales_on_events['sales_foods']-no_event_sales_cat
18 cat_sales_on_events['sales_hobbies_diff']=cat_sales_on_events['sales_hobbies'
19 cat_sales_on_events['sales_household_diff']=cat_sales_on_events['sales_household'
20 fig=go.Figure()
21 fig.add_trace(go.Bar(x=cat_sales_on_events['sales_foods_diff'],y=cat_sales_on_events
22 fig.add_trace(go.Bar(x=cat_sales_on_events['sales_hobbies_diff'],y=cat_sales_on_events
23 fig.add_trace(go.Bar(x=cat_sales_on_events['sales_household_diff'],y=cat_sales_on_events
24 fig.update_layout(width=700,
25 height=500,

16 of 24 11/8/2021, 2:32 PM
WALMART UNIT SALES PREDICTION — A Time Series ... https://m-mazumdar13.medium.com/walmart-unit-sales-predi...

1 state_cat_sales=pd.melt(sales_df.groupby(['state_id','cat_id'],as_index=False
2 id_vars=['state_id','cat_id'],var_name='d',value_name
3
4 state_cat_sales=state_cat_sales.merge(calendar_df[['snap_CA','snap_TX','snap_WI'
5
6 snap_CA_sales=state_cat_sales[state_cat_sales['state_id']=='CA'].groupby(['cat_id'
7 snap_WI_sales=state_cat_sales[state_cat_sales['state_id']=='WI'].groupby(['cat_id'
8 snap_TX_sales=state_cat_sales[state_cat_sales['state_id']=='TX'].groupby(['cat_id'
9
10 fig,axes=plt.subplots(1,3,figsize=(12,5))
11 snap_CA_sales.pivot("cat_id", "snap_CA", "sales").plot(kind='bar',ax=axes[0],
12 snap_WI_sales.pivot("cat_id", "snap_WI", "sales").plot(kind='bar',ax=axes[1])
13 snap_TX_sales pivot("cat_id" "snap_TX" "sales").plot(kind='bar' ax=axes[2])

18 of 24 11/8/2021, 2:32 PM
WALMART UNIT SALES PREDICTION — A Time Series ... https://m-mazumdar13.medium.com/walmart-unit-sales-predi...

1 price_df_1=price_df.merge(sales_df[['item_id','cat_id','dept_id']].drop_duplicates
2 del price_df
3 price_df_final=price_df_1.groupby(['wm_yr_wk','cat_id','dept_id'],as_index=False
4 del price_df_1
5 price_df_final=price_df_final[price_df_final['wm_yr_wk']<=11613]
6
7
8 sales_df_1=sales_df.melt(id_vars=['id','item_id','dept_id','cat_id','store_id'
9 var_name='d',value_name='sales')
10 del sales_df
11 sales_df_2=sales_df_1.merge(calendar_df[['d','date','wm_yr_wk']].drop_duplicates
12 del sales_df_1
13 sales_df_final=sales_df_2.groupby(['wm_yr_wk','cat_id','dept_id'],as_index=False
14
15
16 price_df_final['sales']=sales_df_final['sales']
17
18 for dept,row in zip(price_df_final['dept_id'].unique(),range(1,8)):
19 fig=make_subplots(rows=1,cols=2)
20 sales_trend=seasonal_decompose(price_df_final[price_df_final['dept_id'
21 price_trend=seasonal_decompose(price_df_final[price_df_final['dept_id'
22
23 fig.append_trace(go.Line(x=price_df_final[price_df_final['dept_id']==
24 ,y=sales_trend,name=dept+' weekly average sales trend'),row=1
25
26 fig.append_trace(go.Line(x=price_df_final[price_df_final['dept_id']==
27 price_trend dept ' weekly average sell price trend'),

19 of 24 11/8/2021, 2:32 PM
WALMART UNIT SALES PREDICTION — A Time Series ... https://m-mazumdar13.medium.com/walmart-unit-sales-predi...

20 of 24 11/8/2021, 2:32 PM
WALMART UNIT SALES PREDICTION — A Time Series ... https://m-mazumdar13.medium.com/walmart-unit-sales-predi...

21 of 24 11/8/2021, 2:32 PM
WALMART UNIT SALES PREDICTION — A Time Series ... https://m-mazumdar13.medium.com/walmart-unit-sales-predi...

22 of 24 11/8/2021, 2:32 PM
WALMART UNIT SALES PREDICTION — A Time Series ... https://m-mazumdar13.medium.com/walmart-unit-sales-predi...

23 of 24 11/8/2021, 2:32 PM
WALMART UNIT SALES PREDICTION — A Time Series ... https://m-mazumdar13.medium.com/walmart-unit-sales-predi...

24 of 24 11/8/2021, 2:32 PM