Commit
•
e14c6fe
1
Parent(s):
590ad2e
First commit streamlit application
Browse files- app.py +289 -0
- requirements.txt +7 -0
- static/customer_info.csv +0 -0
- static/logo_artefact.png +0 -0
- static/logo_random.png +0 -0
app.py
ADDED
@@ -0,0 +1,289 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import pandas as pd
|
2 |
+
import numpy as np
|
3 |
+
import matplotlib.pyplot as plt
|
4 |
+
import seaborn as sns
|
5 |
+
from PIL import Image
|
6 |
+
import datetime
|
7 |
+
import streamlit as st
|
8 |
+
|
9 |
+
#####################################################################################################################################
|
10 |
+
st.set_page_config(layout='wide')
|
11 |
+
|
12 |
+
# Sidebar: Image + main info on dataset
|
13 |
+
def data_subset(data, beginning='2010-12-01', end='2011-12-09'):
|
14 |
+
|
15 |
+
beginning = pd.to_datetime(beginning)
|
16 |
+
end = pd.to_datetime(end)
|
17 |
+
|
18 |
+
# Subsetting
|
19 |
+
data = data[(data['InvoiceDate'] >= beginning) & (data['InvoiceDate'] <= end)]
|
20 |
+
|
21 |
+
return data
|
22 |
+
|
23 |
+
# Loading datasets
|
24 |
+
df_info = pd.read_csv('static/customer_info.csv')
|
25 |
+
df_info['InvoiceDate'] = pd.to_datetime(df_info['InvoiceDate'])
|
26 |
+
|
27 |
+
with st.sidebar:
|
28 |
+
col1, col2, col3 = st.columns(3)
|
29 |
+
with col2:
|
30 |
+
random_image = Image.open('static/logo_random.png')
|
31 |
+
st.image(random_image)
|
32 |
+
|
33 |
+
# Showing top products
|
34 |
+
if st.checkbox('Check to see top products sold in a selected timeframe'):
|
35 |
+
start = st.date_input('Input beginning of the wanted timeframe', datetime.date(2010, 12, 1),
|
36 |
+
min_value=datetime.date(2010, 12, 1), max_value=datetime.date(2011, 12, 9), key=1)
|
37 |
+
end = st.date_input('Input beginning of the wanted timeframe', datetime.date(2011, 12, 9),
|
38 |
+
min_value=start, max_value=datetime.date(2011, 12, 9), key=2)
|
39 |
+
|
40 |
+
df_top_products = df_info.copy()
|
41 |
+
df_subset_products = data_subset(df_top_products, start, end)
|
42 |
+
|
43 |
+
df_subset_products = df_top_products.groupby('Description')['Quantity'].sum()
|
44 |
+
number_chosen_products = st.number_input('How many top products sold do you want to see?', value=5)
|
45 |
+
df_subset_products_top = pd.DataFrame(df_top_products.sort_values(by='Quantity', ascending=False)).iloc[:number_chosen_products,:]
|
46 |
+
df_subset_products_top = df_subset_products_top[['Description', 'Quantity']]
|
47 |
+
st.dataframe(df_subset_products_top)
|
48 |
+
|
49 |
+
# Showing most recent clients
|
50 |
+
if st.checkbox('Check to see the most recent customers in a selected timeframe'):
|
51 |
+
start_clts = st.date_input('Input beginning of the wanted timeframe', datetime.date(2010, 12, 1),
|
52 |
+
min_value=datetime.date(2010, 12, 1), max_value=datetime.date(2011, 12, 9), key=3)
|
53 |
+
end_clts = st.date_input('Input beginning of the wanted timeframe', datetime.date(2011, 12, 9),
|
54 |
+
min_value=start_clts, max_value=datetime.date(2011, 12, 9), key=4)
|
55 |
+
df_recent_customers = df_info.copy()
|
56 |
+
df_subset_recent_customers = data_subset(df_recent_customers, start_clts, end_clts)
|
57 |
+
|
58 |
+
df_subset_recent_customers = df_subset_recent_customers.groupby('CustomerID')['Recency'].min()
|
59 |
+
number_chosen_recency = st.number_input('How many recent customers do you want to see?', value=5)
|
60 |
+
df_subset_recent_customers_top = pd.DataFrame(df_subset_recent_customers.sort_values()).iloc[:number_chosen_recency,:]
|
61 |
+
st.dataframe(df_subset_recent_customers_top)
|
62 |
+
|
63 |
+
# Showing most prolific customers
|
64 |
+
if st.checkbox('Check to see the top customers in a selected timeframe'):
|
65 |
+
start_top = st.date_input('Input beginning of the wanted timeframe', datetime.date(2010, 12, 1),
|
66 |
+
min_value=datetime.date(2010, 12, 1), max_value=datetime.date(2011, 12, 9), key=5)
|
67 |
+
end_top = st.date_input('Input beginning of the wanted timeframe', datetime.date(2011, 12, 9),
|
68 |
+
min_value=start_top, max_value=datetime.date(2011, 12, 9), key=6)
|
69 |
+
df_top_customers = df_info.copy()
|
70 |
+
df_subset_top_customers = data_subset(df_top_customers, start_top, end_top)
|
71 |
+
|
72 |
+
df_subset_top_customers = df_subset_top_customers.groupby('CustomerID')['Monetary'].sum()
|
73 |
+
number_chosen_top_clts = st.number_input('How many top customers do you want to see?', value=5)
|
74 |
+
df_subset_top_customers_top = pd.DataFrame(df_subset_top_customers.sort_values(ascending=False)).iloc[:number_chosen_top_clts,:]
|
75 |
+
st.dataframe(df_subset_top_customers_top)
|
76 |
+
|
77 |
+
#####################################################################################################################################
|
78 |
+
st.title('E-commerce: client dashboard')
|
79 |
+
st.write("---")
|
80 |
+
|
81 |
+
# Loading dataset
|
82 |
+
df_info_customer = df_info.copy()
|
83 |
+
customer_id_default = int(df_info_customer['CustomerID'].min())
|
84 |
+
|
85 |
+
# We choose a CustomerID
|
86 |
+
st.number_input('CustomerID', min_value=customer_id_default, value=customer_id_default, step=1, format="%d", key='customer_id')
|
87 |
+
customer_id = st.session_state.customer_id
|
88 |
+
|
89 |
+
if customer_id not in df_info_customer['CustomerID'].values:
|
90 |
+
st.write('This CustomerID is not available right now, please find another.')
|
91 |
+
|
92 |
+
else:
|
93 |
+
start_info = st.date_input('Input beginning of the wanted timeframe', datetime.date(2010, 12, 1),
|
94 |
+
min_value=datetime.date(2010, 12, 1), max_value=datetime.date(2011, 12, 9), key=7)
|
95 |
+
end_info = st.date_input('Input beginning of the wanted timeframe', datetime.date(2011, 12, 9),
|
96 |
+
min_value=start_info, max_value=datetime.date(2011, 12, 9), key=8)
|
97 |
+
df_subset_info_customer = data_subset(df_info_customer, start_info, end_info)
|
98 |
+
|
99 |
+
# Main info (recency, number of orders, how much the customer spent)
|
100 |
+
df_subset_info_customer = df_subset_info_customer[df_subset_info_customer['CustomerID'] == customer_id]
|
101 |
+
df_main_info = df_subset_info_customer.groupby('CustomerID').agg(Recency=('Recency', 'min'), NbOrder=('NbOrder', 'max'), MonetaryTotal=('Monetary', 'sum'))
|
102 |
+
|
103 |
+
# GroupBy to get the mean value of each order for the customer
|
104 |
+
df_mean_order = df_subset_info_customer.groupby(['InvoiceNo', 'CustomerID']).agg(TotalOrderValue=('Monetary', 'sum'))
|
105 |
+
df_mean_order = df_mean_order.groupby('CustomerID').agg(MeanOrderValue=('TotalOrderValue', 'mean'))
|
106 |
+
|
107 |
+
# GroupBy to get the most bought product and its quantity
|
108 |
+
df_product_clts = pd.DataFrame(df_info.groupby(['CustomerID','Description'])['Quantity'].sum())
|
109 |
+
df_product_clts = df_product_clts.reset_index()
|
110 |
+
df_product_clts = df_product_clts[df_product_clts['CustomerID'] == customer_id]
|
111 |
+
ids, values = df_product_clts.groupby('CustomerID')['Quantity'].max().index, df_product_clts.groupby('CustomerID')['Quantity'].max().values
|
112 |
+
df_product_clts = df_product_clts[(df_product_clts['CustomerID'] == ids[0]) & (df_product_clts['Quantity'] == values[0])]
|
113 |
+
|
114 |
+
# Now we create the columns we want
|
115 |
+
df_main_info['MeanOrderValue'] = df_mean_order['MeanOrderValue'].values[0]
|
116 |
+
df_main_info['MostOrderedProduct'] = df_product_clts['Description'].values[0]
|
117 |
+
df_main_info['MostOrderedProductQuantity'] = df_product_clts['Quantity'].values[0]
|
118 |
+
|
119 |
+
# We can show it now that it's complete
|
120 |
+
st.dataframe(df_main_info)
|
121 |
+
|
122 |
+
st.write("---")
|
123 |
+
#####################################################################################################################################
|
124 |
+
st.subheader('Similarity between customers:')
|
125 |
+
with st.expander('Choose a number of similar customers to compare:'):
|
126 |
+
|
127 |
+
if st.checkbox('Only similar customers:'):
|
128 |
+
options_similar = ['Recency', 'NbOrder', 'MonetaryTotal', 'MeanOrderValue']
|
129 |
+
option_similar = st.selectbox('Choose a feature to plot:', tuple(options_similar))
|
130 |
+
|
131 |
+
df_similar_customer = df_info.copy()
|
132 |
+
|
133 |
+
# Main info (recency, number of orders, how much the customer spent)
|
134 |
+
df_similar_customer_grouped = df_similar_customer.groupby('CustomerID').agg(Recency=('Recency', 'min'), NbOrder=('NbOrder', 'max'), MonetaryTotal=('Monetary', 'sum'))
|
135 |
+
|
136 |
+
# GroupBy to get the mean value of each order for the customer
|
137 |
+
df_mean_order_similar = df_similar_customer.groupby(['InvoiceNo', 'CustomerID']).agg(TotalOrderValue=('Monetary', 'sum'))
|
138 |
+
df_mean_order_similar = df_mean_order_similar.groupby('CustomerID').agg(MeanOrderValue=('TotalOrderValue', 'mean'))
|
139 |
+
|
140 |
+
# Now we create the column we want
|
141 |
+
df_similar_customer_grouped['MeanOrderValue'] = df_mean_order_similar['MeanOrderValue'].values
|
142 |
+
|
143 |
+
# We select the client
|
144 |
+
df_similar_customer_grouped = df_similar_customer_grouped.reset_index()
|
145 |
+
df_selected_clt = df_similar_customer_grouped[df_similar_customer_grouped['CustomerID'] == customer_id]
|
146 |
+
|
147 |
+
# We calculate distances (euclidean)
|
148 |
+
distances = []
|
149 |
+
for i in range(df_similar_customer_grouped.shape[0]):
|
150 |
+
distance = np.linalg.norm(df_similar_customer_grouped.drop('CustomerID', axis=1).values[i] - df_selected_clt.drop('CustomerID', axis=1).values)
|
151 |
+
distances.append(distance)
|
152 |
+
|
153 |
+
n_neighbors = st.slider("Number of similar customers:", min_value=5, max_value=30, value=10)
|
154 |
+
neighbors = sorted(distances)[:n_neighbors]
|
155 |
+
|
156 |
+
# We get the indices of the similar customers
|
157 |
+
indices_neighbors = []
|
158 |
+
for i in range(len(neighbors)):
|
159 |
+
indices_neighbors.append(distances.index(neighbors[i]))
|
160 |
+
|
161 |
+
df_neighbors_selected = df_similar_customer_grouped.iloc[indices_neighbors, :]
|
162 |
+
|
163 |
+
fig2, ax = plt.subplots()
|
164 |
+
ax.set_xlabel('Customers', fontsize=17)
|
165 |
+
ax.set_ylabel(option_similar, fontsize=17)
|
166 |
+
ax.axhline(y=df_selected_clt[option_similar].values, color='r', label='axhline - full height')
|
167 |
+
ax = plt.boxplot(df_neighbors_selected[option_similar], showfliers=False)
|
168 |
+
|
169 |
+
st.pyplot(fig2)
|
170 |
+
|
171 |
+
if st.checkbox('Compare to all customers:'):
|
172 |
+
options_all = ['Recency', 'NbOrder', 'MonetaryTotal', 'MeanOrderValue']
|
173 |
+
option_all = st.selectbox('Choose a feature to plot:', tuple(options_all))
|
174 |
+
|
175 |
+
df_all_customer = df_info.copy()
|
176 |
+
|
177 |
+
# Main info (recency, number of orders, how much the customer spent)
|
178 |
+
df_all_customer_grouped = df_all_customer.groupby('CustomerID').agg(Recency=('Recency', 'min'), NbOrder=('NbOrder', 'max'), MonetaryTotal=('Monetary', 'sum'))
|
179 |
+
|
180 |
+
# GroupBy to get the mean value of each order for the customer
|
181 |
+
df_mean_order_all = df_all_customer.groupby(['InvoiceNo', 'CustomerID']).agg(TotalOrderValue=('Monetary', 'sum'))
|
182 |
+
df_mean_order_all = df_mean_order_all.groupby('CustomerID').agg(MeanOrderValue=('TotalOrderValue', 'mean'))
|
183 |
+
|
184 |
+
# Now we create the column we want
|
185 |
+
df_all_customer_grouped['MeanOrderValue'] = df_mean_order_all['MeanOrderValue'].values
|
186 |
+
|
187 |
+
# We select the client
|
188 |
+
df_selected_clt_all = df_all_customer_grouped.reset_index()
|
189 |
+
df_selected_clt_all = df_selected_clt_all[df_selected_clt_all['CustomerID'] == customer_id]
|
190 |
+
|
191 |
+
# We calculate distances (euclidean)
|
192 |
+
distances = []
|
193 |
+
for i in range(df_all_customer_grouped.shape[0]):
|
194 |
+
distance = np.linalg.norm(df_all_customer_grouped.values[i] - df_selected_clt_all.drop('CustomerID', axis=1).values)
|
195 |
+
distances.append(distance)
|
196 |
+
|
197 |
+
fig2, ax = plt.subplots()
|
198 |
+
ax.set_xlabel('Customers', fontsize=17)
|
199 |
+
ax.set_ylabel(option_all, fontsize=17)
|
200 |
+
ax.axhline(y=df_selected_clt_all[option_all].values, color='r', label='axhline - full height')
|
201 |
+
ax = plt.boxplot(df_all_customer_grouped[option_all], showfliers=False)
|
202 |
+
|
203 |
+
st.pyplot(fig2)
|
204 |
+
|
205 |
+
st.write("---")
|
206 |
+
#####################################################################################################################################
|
207 |
+
st.subheader('Barplot of top selected products in the selected timeframe:')
|
208 |
+
with st.expander('Select to choose how many top products you want to see and in which timeframe'):
|
209 |
+
|
210 |
+
start_product_date = st.date_input('Input beginning of the wanted timeframe', datetime.date(2010, 12, 1),
|
211 |
+
min_value=datetime.date(2010, 12, 1), max_value=datetime.date(2011, 12, 9), key=9)
|
212 |
+
end_product_date = st.date_input('Input beginning of the wanted timeframe', datetime.date(2011, 12, 9),
|
213 |
+
min_value=start_product_date, max_value=datetime.date(2011, 12, 9), key=10)
|
214 |
+
df_top_products_plot = df_info.copy()
|
215 |
+
df_subset_products = data_subset(df_top_products_plot, start_product_date, end_product_date)
|
216 |
+
start_product, end_product = st.select_slider('Select a range of top product', options=[x for x in range(1, 21)], value=(1, 10))
|
217 |
+
df_subset_products = df_subset_products.groupby('Description')['Quantity'].sum()
|
218 |
+
df_subset_products = df_subset_products.reset_index()
|
219 |
+
df_slider_products = df_subset_products.sort_values(by='Quantity', ascending=False)
|
220 |
+
df_slider_products = df_slider_products.iloc[start_product-1:end_product, :]
|
221 |
+
|
222 |
+
fig, ax = plt.subplots()
|
223 |
+
bars = plt.barh(y=df_slider_products['Description'], width=df_slider_products['Quantity'], color=['darkmagenta', 'darkblue', 'darkgreen', 'darkred', 'darkgrey', 'darkorange'])
|
224 |
+
|
225 |
+
ax.bar_label(bars)
|
226 |
+
ax = plt.gca().invert_yaxis()
|
227 |
+
|
228 |
+
st.subheader('Selected top products:')
|
229 |
+
st.pyplot(fig)
|
230 |
+
|
231 |
+
st.write("---")
|
232 |
+
#####################################################################################################################################
|
233 |
+
|
234 |
+
st.subheader('Barplot of sales:')
|
235 |
+
with st.expander('Select to choose the periodicity:'):
|
236 |
+
options_similar = ['Months', 'Days', 'Hours']
|
237 |
+
option_similar = st.selectbox('Choose a periodicity:', tuple(options_similar))
|
238 |
+
|
239 |
+
if option_similar == 'Months':
|
240 |
+
df_months = df_info.copy()
|
241 |
+
df_months = df_months.merge(pd.DataFrame(df_months.groupby('CustomerID')['Monetary'].sum()), on='CustomerID')
|
242 |
+
df_months['Periodicity'] = pd.DatetimeIndex(df_months['InvoiceDate']).month
|
243 |
+
df_months = df_months.sort_values('Recency')
|
244 |
+
df_months = df_months.drop_duplicates(subset='CustomerID')
|
245 |
+
|
246 |
+
fig1, ax1 = plt.subplots()
|
247 |
+
ax1 = sns.barplot(x=df_months['Periodicity'], y=df_months['Monetary_y'], errorbar=None)
|
248 |
+
plt.title('Sales per Months')
|
249 |
+
plt.xlabel('Periodicity: Months')
|
250 |
+
plt.ylabel('TotalOrderValue')
|
251 |
+
st.pyplot(fig1)
|
252 |
+
|
253 |
+
elif option_similar == 'Days':
|
254 |
+
df_days = df_info.copy()
|
255 |
+
df_days = df_days.merge(pd.DataFrame(df_days.groupby('CustomerID')['Monetary'].sum()), on='CustomerID')
|
256 |
+
df_days['Periodicity'] = pd.DatetimeIndex(df_days['InvoiceDate']).day
|
257 |
+
df_days = df_days.sort_values('Recency')
|
258 |
+
df_days = df_days.drop_duplicates(subset='CustomerID')
|
259 |
+
|
260 |
+
fig2, ax2 = plt.subplots()
|
261 |
+
ax2 = sns.barplot(x=df_days['Periodicity'], y=df_days['Monetary_y'], errorbar=None)
|
262 |
+
plt.title('Sales per Days')
|
263 |
+
plt.xlabel('Periodicity: Days')
|
264 |
+
plt.xticks(rotation=90)
|
265 |
+
plt.ylabel('TotalOrderValue')
|
266 |
+
st.pyplot(fig2)
|
267 |
+
|
268 |
+
elif option_similar == 'Hours':
|
269 |
+
df_hours = df_info.copy()
|
270 |
+
df_hours = df_hours.merge(pd.DataFrame(df_hours.groupby('CustomerID')['Monetary'].sum()), on='CustomerID')
|
271 |
+
df_hours['Periodicity'] = pd.DatetimeIndex(df_hours['InvoiceDate']).hour
|
272 |
+
df_hours = df_hours.sort_values('Recency')
|
273 |
+
df_hours = df_hours.drop_duplicates(subset='CustomerID')
|
274 |
+
|
275 |
+
fig3, ax3 = plt.subplots()
|
276 |
+
ax3 = sns.barplot(x=df_hours['Periodicity'], y=df_hours['Monetary_y'], errorbar=None)
|
277 |
+
plt.title('Sales per Hours')
|
278 |
+
plt.xlabel('Periodicity: Hours')
|
279 |
+
plt.ylabel('TotalOrderValue')
|
280 |
+
st.pyplot(fig3)
|
281 |
+
|
282 |
+
|
283 |
+
st.write("---")
|
284 |
+
#####################################################################################################################################
|
285 |
+
|
286 |
+
col1, col2, col3, col4, col5 = st.columns(5)
|
287 |
+
with col5:
|
288 |
+
logo_artefact = Image.open('static/logo_artefact.png')
|
289 |
+
st.image(logo_artefact)
|
requirements.txt
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
streamlit
|
2 |
+
pandas
|
3 |
+
numpy
|
4 |
+
matplotlib
|
5 |
+
datetime
|
6 |
+
Pillow
|
7 |
+
plotly
|
static/customer_info.csv
ADDED
The diff for this file is too large to render.
See raw diff
|
|
static/logo_artefact.png
ADDED
![]() |
static/logo_random.png
ADDED
![]() |