Machine Learning: Using Python to Find Your Next Business Opportunity¶

Inspired by the course "Advanced Python Projects: Build AI Applications" by Priya Mohan on LinkedIn Learning, this project delves into AI development using Python. Throughout this work, we explore advanced AI concepts, techniques, and applications to elevate our Python portfolio and enhance our understanding of AI development.

Mohan, Priya. "Advanced Python Projects: Build AI Applications." LinkedIn Learning, 2024.

In [2]:
!pip install pandas scikit-learn matplotlib
!pip install alpha_vantage pandas

Requirement already satisfied: pandas in c:\users\nayel\anaconda3\lib\site-packages (2.2.1)
Requirement already satisfied: scikit-learn in c:\users\nayel\anaconda3\lib\site-packages (1.2.2)
Requirement already satisfied: matplotlib in c:\users\nayel\anaconda3\lib\site-packages (3.8.0)
Requirement already satisfied: numpy<2,>=1.23.2 in c:\users\nayel\anaconda3\lib\site-packages (from pandas) (1.26.4)
Requirement already satisfied: python-dateutil>=2.8.2 in c:\users\nayel\anaconda3\lib\site-packages (from pandas) (2.8.2)
Requirement already satisfied: pytz>=2020.1 in c:\users\nayel\anaconda3\lib\site-packages (from pandas) (2023.3.post1)
Requirement already satisfied: tzdata>=2022.7 in c:\users\nayel\anaconda3\lib\site-packages (from pandas) (2023.3)
Requirement already satisfied: scipy>=1.3.2 in c:\users\nayel\anaconda3\lib\site-packages (from scikit-learn) (1.11.4)
Requirement already satisfied: joblib>=1.1.1 in c:\users\nayel\anaconda3\lib\site-packages (from scikit-learn) (1.2.0)
Requirement already satisfied: threadpoolctl>=2.0.0 in c:\users\nayel\anaconda3\lib\site-packages (from scikit-learn) (2.2.0)
Requirement already satisfied: contourpy>=1.0.1 in c:\users\nayel\anaconda3\lib\site-packages (from matplotlib) (1.2.0)
Requirement already satisfied: cycler>=0.10 in c:\users\nayel\anaconda3\lib\site-packages (from matplotlib) (0.11.0)
Requirement already satisfied: fonttools>=4.22.0 in c:\users\nayel\anaconda3\lib\site-packages (from matplotlib) (4.25.0)
Requirement already satisfied: kiwisolver>=1.0.1 in c:\users\nayel\anaconda3\lib\site-packages (from matplotlib) (1.4.4)
Requirement already satisfied: packaging>=20.0 in c:\users\nayel\anaconda3\lib\site-packages (from matplotlib) (23.1)
Requirement already satisfied: pillow>=6.2.0 in c:\users\nayel\anaconda3\lib\site-packages (from matplotlib) (10.2.0)
Requirement already satisfied: pyparsing>=2.3.1 in c:\users\nayel\anaconda3\lib\site-packages (from matplotlib) (3.0.9)
Requirement already satisfied: six>=1.5 in c:\users\nayel\anaconda3\lib\site-packages (from python-dateutil>=2.8.2->pandas) (1.16.0)
Requirement already satisfied: alpha_vantage in c:\users\nayel\anaconda3\lib\site-packages (2.3.1)
Requirement already satisfied: pandas in c:\users\nayel\anaconda3\lib\site-packages (2.2.1)
Requirement already satisfied: aiohttp in c:\users\nayel\anaconda3\lib\site-packages (from alpha_vantage) (3.9.3)
Requirement already satisfied: requests in c:\users\nayel\anaconda3\lib\site-packages (from alpha_vantage) (2.31.0)
Requirement already satisfied: numpy<2,>=1.23.2 in c:\users\nayel\anaconda3\lib\site-packages (from pandas) (1.26.4)
Requirement already satisfied: python-dateutil>=2.8.2 in c:\users\nayel\anaconda3\lib\site-packages (from pandas) (2.8.2)
Requirement already satisfied: pytz>=2020.1 in c:\users\nayel\anaconda3\lib\site-packages (from pandas) (2023.3.post1)
Requirement already satisfied: tzdata>=2022.7 in c:\users\nayel\anaconda3\lib\site-packages (from pandas) (2023.3)
Requirement already satisfied: six>=1.5 in c:\users\nayel\anaconda3\lib\site-packages (from python-dateutil>=2.8.2->pandas) (1.16.0)
Requirement already satisfied: aiosignal>=1.1.2 in c:\users\nayel\anaconda3\lib\site-packages (from aiohttp->alpha_vantage) (1.2.0)
Requirement already satisfied: attrs>=17.3.0 in c:\users\nayel\anaconda3\lib\site-packages (from aiohttp->alpha_vantage) (23.1.0)
Requirement already satisfied: frozenlist>=1.1.1 in c:\users\nayel\anaconda3\lib\site-packages (from aiohttp->alpha_vantage) (1.4.0)
Requirement already satisfied: multidict<7.0,>=4.5 in c:\users\nayel\anaconda3\lib\site-packages (from aiohttp->alpha_vantage) (6.0.4)
Requirement already satisfied: yarl<2.0,>=1.0 in c:\users\nayel\anaconda3\lib\site-packages (from aiohttp->alpha_vantage) (1.9.3)
Requirement already satisfied: charset-normalizer<4,>=2 in c:\users\nayel\anaconda3\lib\site-packages (from requests->alpha_vantage) (2.0.4)
Requirement already satisfied: idna<4,>=2.5 in c:\users\nayel\anaconda3\lib\site-packages (from requests->alpha_vantage) (3.4)
Requirement already satisfied: urllib3<3,>=1.21.1 in c:\users\nayel\anaconda3\lib\site-packages (from requests->alpha_vantage) (2.0.7)
Requirement already satisfied: certifi>=2017.4.17 in c:\users\nayel\anaconda3\lib\site-packages (from requests->alpha_vantage) (2024.6.2)
In [3]:
#import all of required libraries and classes right here
import pandas as pd
import re
from sklearn.preprocessing import StandardScaler
import matplotlib.pyplot as plt
import numpy as np
import warnings
from sklearn.metrics import mean_squared_error as mse
from sklearn.model_selection import train_test_split, GridSearchCV
from sklearn.ensemble import RandomForestRegressor, GradientBoostingRegressor
from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_absolute_error, mean_squared_error, r2_score
In [4]:
# Data Ingestion and Analysis
df= pd.read_excel('Coffee_shop_data.xlsx')
population = pd.read_csv('population.csv', skiprows=[0])
In [5]:
population.head()
Out[5]:
Geography Label for GEO_ID Race/Ethnic Group Population Groups Total Errata of Total Total!!1-person household Total!!2-person household Total!!3-person household Total!!4-person household Total!!5-person household Total!!6-person household Total!!7-or-more-person household
0 8600000US89010 ZCTA5 89010 1 Total population 172 NaN 51 69 19 12 12 4 5
1 8600000US89019 ZCTA5 89019 1 Total population 1164 NaN 412 421 134 83 57 32 25
2 8600000US89060 ZCTA5 89060 1 Total population 4144 NaN 1106 1714 550 361 222 95 96
3 8600000US89061 ZCTA5 89061 1 Total population 2109 NaN 469 936 283 206 112 62 41
4 8600000US89439 ZCTA5 89439 1 Total population 671 NaN 193 314 88 55 17 2 2
In [6]:
df.head()
Out[6]:
ID no. Business Name Street address City State Zip Code Phone Rating Gender majority Median Salary Latte Price
0 1 Brew Haven 8 Old Shore Place Oakland California 94616 415-810-4769 4.5 Male 72463 3.31000
1 2 Bean Bliss 6650 Clarendon Crossing Stockton California 95210 209-701-1665 5.0 Female 87117 5.35000
2 3 Caffeine Cove 7281 Buell Road Fresno California 93773 559-137-3554 4.7 Male 86394 4.15000
3 4 Mug Magic 670 Jackson Avenue Torrance California 90510 818-789-5573 4.3 Male 88343 5.34000
4 5 Daily Grind 37 Ludington Terrace San Diego California 92196 619-354-2389 1.6 Male 77795 4.51175
In [7]:
# check for data info
df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 1000 entries, 0 to 999
Data columns (total 11 columns):
 #   Column           Non-Null Count  Dtype  
---  ------           --------------  -----  
 0   ID no.           1000 non-null   int64  
 1   Business Name    1000 non-null   object 
 2   Street address   1000 non-null   object 
 3   City             1000 non-null   object 
 4   State            1000 non-null   object 
 5   Zip Code         1000 non-null   int64  
 6   Phone            1000 non-null   object 
 7   Rating           1000 non-null   float64
 8   Gender majority  1000 non-null   object 
 9   Median Salary    1000 non-null   int64  
 10  Latte Price      1000 non-null   float64
dtypes: float64(2), int64(3), object(6)
memory usage: 86.1+ KB
In [8]:
df.shape
Out[8]:
(1000, 11)
In [9]:
population.shape
Out[9]:
(1705, 13)
In [10]:
#get basic stats about the data
df.describe()
Out[10]:
ID no. Zip Code Rating Median Salary Latte Price
count 1000.000000 1000.000000 1000.000000 1000.000000 1000.000000
mean 500.500000 92976.163000 3.784600 81182.842000 5.061491
std 288.819436 1706.943177 1.150717 5142.670356 0.352002
min 1.000000 90005.000000 1.000000 72001.000000 3.090000
25% 250.750000 91751.750000 3.200000 76776.000000 4.830175
50% 500.500000 92883.000000 4.240000 81113.000000 5.062450
75% 750.250000 94257.000000 4.580000 85684.000000 5.310025
max 1000.000000 96154.000000 5.000000 89978.000000 6.480000
In [11]:
ax=df['City'].value_counts().head(5).plot(kind='bar')
ax.set_title('Top 5 cities with most coffee shops')
plt.show()
No description has been provided for this image
In [12]:
#the top 10 mosr famous brands
ax=df['Business Name'].value_counts().head(10).plot(kind='bar')
ax.set_title('Top most famous brands')
plt.show()
No description has been provided for this image

Data Preprocessing¶

In [13]:
#if we have null values we would impute it. Missing values -replace it with the mode (most occuring values)
df.isna().sum()
Out[13]:
ID no.             0
Business Name      0
Street address     0
City               0
State              0
Zip Code           0
Phone              0
Rating             0
Gender majority    0
Median Salary      0
Latte Price        0
dtype: int64
In [14]:
# converting zipcode to object data (str) in order to join the zip code with the population data
df['Zip Code']=df['Zip Code'].astype(str)
In [15]:
#extract zip code from population
#Getting the last 5 digits from the population zip code. Creating a new column called zip code
def find_zip_code(geocode):
    pattern = r'\d{5}$'
    
    match = re.search(pattern, geocode)
    
    if match:
        zip_code = match.group(0)
        return zip_code
In [16]:
#the actual conversion is below. The above is the function
population['Zip Code']=population['Geography'].apply(find_zip_code)
In [17]:
#merging the population via zip code as population is an important feature to determinate the price / locations 
cafe_data=df.copy()
#notice that the data size is reduce after a join
df=pd.merge(cafe_data,population)
In [18]:
#keeping only Total from population.In the pop dataset, keeoing total population column and other columns.
columns=cafe_data.columns.values.tolist()+['Total']
df=df[columns]
#rename Total to population
df=df.rename(columns={"Total":"Population"})
In [19]:
df
Out[19]:
ID no. Business Name Street address City State Zip Code Phone Rating Gender majority Median Salary Latte Price Population
0 2 Bean Bliss 6650 Clarendon Crossing Stockton California 95210 209-701-1665 5.00 Female 87117 5.35000 11180
1 6 Espresso Elegance 8427 Atwood Road Visalia California 93291 559-929-4731 1.20 Female 78753 5.31765 15310
2 10 Perk Palace 7 David Junction Fresno California 93726 559-323-2365 4.20 Female 80973 5.30665 13942
3 11 The Coffee Cartel 6018 Rockefeller Center Orange California 92867 714-292-8390 1.90 Male 74340 4.43900 13284
4 14 Grindhouse Cafe 7846 Cherokee Junction Visalia California 93291 559-253-9426 4.50 Male 78753 5.27165 15310
... ... ... ... ... ... ... ... ... ... ... ... ...
407 982 Perk Paradise 4 Park Meadow Crossing Van Nuys California 91411 818-584-8823 4.00 Female 76034 4.99170 9177
408 983 Urban Brewtropolis 2459 Golden Leaf Place Santa Monica California 90405 818-215-2671 4.82 Female 86920 4.98200 14376
409 985 Caffeine Communion 40 Sommers Parkway San Jose California 95133 408-304-3646 4.04 Female 78176 4.64680 7365
410 998 Urban Brewtropolis 141 Burrows Place Los Angeles California 90035 323-785-4094 3.00 Male 87604 4.64000 12814
411 999 Steamy Moments 42004 Bellgrove Terrace Orange California 92867 714-506-9394 4.50 Male 74340 3.82000 13284

412 rows × 12 columns

In [20]:
#Keeping only relevant features
df=df[['Zip Code', 'Rating', 'Median Salary', 'Latte Price', 'Population']]
df.shape
Out[20]:
(412, 5)
In [21]:
df.columns
Out[21]:
Index(['Zip Code', 'Rating', 'Median Salary', 'Latte Price', 'Population'], dtype='object')
In [22]:
# Calculate the total number of coffee shops for each zip code
coffee_shop_counts = df['Zip Code'].value_counts().reset_index()
coffee_shop_counts.columns = ['Zip Code', 'CoffeeShopCount']

# Ensure 'Zip Code' is of type string in both DataFrames
df.loc[:, 'Zip Code'] = df['Zip Code'].astype(str)
coffee_shop_counts['Zip Code'] = coffee_shop_counts['Zip Code'].astype(str)

# Merge the counts back into the original DataFrame
df = df.merge(coffee_shop_counts, on='Zip Code', how='left', suffixes=('', '_coffee'))

# Print the updated DataFrame
print(df)

# Criteria:
# a. High population
# b. Low total number of coffee shops
# c. Low ratings
# d. High median salary

# Sorting the DataFrame based on the criteria
sorted_df = df.sort_values(by=['Population', 'CoffeeShopCount', 'Rating', 'Median Salary'],
                           ascending=[False, True, True, False]).reset_index(drop=True)

    Zip Code  Rating  Median Salary  Latte Price  Population  CoffeeShopCount
0      95210    5.00          87117      5.35000       11180                4
1      93291    1.20          78753      5.31765       15310                7
2      93726    4.20          80973      5.30665       13942                5
3      92867    1.90          74340      4.43900       13284                9
4      93291    4.50          78753      5.27165       15310                7
..       ...     ...            ...          ...         ...              ...
407    91411    4.00          76034      4.99170        9177                8
408    90405    4.82          86920      4.98200       14376                3
409    95133    4.04          78176      4.64680        7365                6
410    90035    3.00          87604      4.64000       12814                5
411    92867    4.50          74340      3.82000       13284                9

[412 rows x 6 columns]
In [22]:
# Created a list - if length of list 5, if the zip code is already present, it will not add that into the list. 
# Deduping zip code column and displaying all of the records for the top 5.
lst=[]
for i in range(len(sorted_df)):
    if len(lst)!=5:
        if (sorted_df['Zip Code'][i]) not in lst:
            lst.append(sorted_df['Zip Code'][i])
            
# Filter 'sorted_df' to include only rows where 'Zip Code' is in 'lst'
top_5_zip_codes_df = sorted_df[sorted_df['Zip Code'].isin(lst)]

top_5_zip_codes_df
Out[22]:
Zip Code Rating Median Salary Latte Price Population CoffeeShopCount
0 94110 2.22 74020 5.04500 27128 5
1 94110 3.50 74020 5.04700 27128 5
2 94110 4.40 74020 5.05900 27128 5
3 94110 4.50 74020 4.77900 27128 5
4 94110 4.70 74020 4.58100 27128 5
5 90805 3.00 88140 5.47900 26056 4
6 90805 3.89 88140 5.31900 26056 4
7 90805 4.31 88140 5.02100 26056 4
8 90805 4.37 88140 5.20700 26056 4
9 95823 4.30 73165 4.96825 22470 3
10 95823 4.82 73165 4.87225 22470 3
11 95823 4.94 73165 4.47225 22470 3
12 94544 3.98 72092 4.94860 21872 6
13 94544 4.12 72092 4.96060 21872 6
14 94544 4.17 72092 4.79060 21872 6
15 94544 4.24 72092 4.64460 21872 6
16 94544 4.24 72092 4.91660 21872 6
17 94544 4.86 72092 4.59060 21872 6
18 90025 4.50 85001 5.37405 21228 1
In [23]:
# Features excluding 'Latte Price' and 'Zip Code'
X = df.drop(['Latte Price', 'Zip Code'], axis=1)  
# Target variable
y = df['Latte Price']
In [24]:
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

Scaling¶

In [25]:
sc = StandardScaler()
X_train = sc.fit_transform(X_train)
X_test=sc.transform(X_test)

Model Selection¶

In [26]:
#Model selection
models = {
    'Linear Regression':LinearRegression(),
    'Random Forest': RandomForestRegressor(),
    'Gradient Boosting': GradientBoostingRegressor(),
}

Hyperparameter Tuning¶

In [27]:
param_grid = {
    'Random Forest':{'n_estimators': [50, 100, 200], 'max_depth': [None, 10, 20]},
    'Gradient Boosting': {'n_estimators': [50, 100, 200], 'learning_rate': [0.01, 0.1, 0.2], 'max_depth': [3, 5, 10]},
}
In [28]:
# Perform hyperparameter tuning using GridSearchCV
for model_name, model in models.items():
    if model_name in param_grid:
        grid_search = GridSearchCV(model, param_grid[model_name], cv=5, scoring='neg_mean_squared_error')
        grid_search.fit(X, y)
        
        #Set the best hyperparameters to the model
        models[model_name] = grid_search.best_estimator_

Model Training and Evaluation¶

In [29]:
#Modeling training
for model_name, model in models.items():
    model.fit(X_train, y_train)
In [30]:
#Model Evaluation
for model_name, model in models.items():
    #Evaluate the model on the testing set
    y_pred = model.predict(X_test)
    print(f"{model_name} Metrics:")
    print("Mean Absolute Error:", mean_absolute_error(y_test, y_pred))
    print("Mean Squared Error", mean_squared_error(y_test, y_pred))
    print("R-squared:", r2_score(y_test, y_pred))
    
    print()

Linear Regression Metrics:
Mean Absolute Error: 0.21737899529390042
Mean Squared Error 0.06581667796516916
R-squared: 0.44342320694160287

Random Forest Metrics:
Mean Absolute Error: 0.23160748734464295
Mean Squared Error 0.07573996171310597
R-squared: 0.359507251050331

Gradient Boosting Metrics:
Mean Absolute Error: 0.2184404032308931
Mean Squared Error 0.06691055543719977
R-squared: 0.43417286441134295

Predictions¶

In [31]:
#We want this dataframe to be same as the training data so that model can predict the value
zip_codes_df= top_5_zip_codes_df.drop(['Zip Code', 'Latte Price'], axis=1)
zip_codes_df= sc.transform(zip_codes_df)
In [32]:
for model_name, model in models.items():
    #Predict the prices for lattes in the top 5 zip codes
    predicted_prices = model.predict(zip_codes_df)
In [33]:
predictions = {}

for model_name, model in models.items():
    # Predict the prices for lattes in the top 5 zip codes
    predicted_prices = model.predict(zip_codes_df)
    predictions[model_name] = predicted_prices

# Convert the predictions dictionary to a DataFrame
predictions_df = pd.DataFrame(predictions)
# Add the zip codes to the predictions DataFrame
predictions_df['Zip Code'] = top_5_zip_codes_df['Zip Code'].values

# Rearrange the columns to have 'Zip Code' as the first column
cols = ['Zip Code'] + [col for col in predictions_df.columns if col != 'Zip Code']
predictions_df = predictions_df[cols]

predictions_df
Out[33]:
Zip Code Linear Regression Random Forest Gradient Boosting
0 94110 4.785633 4.999094 4.909164
1 94110 4.781253 5.006316 4.909164
2 94110 4.778173 4.959572 4.909164
3 94110 4.777830 4.965996 4.909164
4 94110 4.777146 4.989925 4.909164
5 90805 5.379973 5.424480 5.304014
6 90805 5.376927 5.364141 5.292189
7 90805 5.375490 5.301565 5.276940
8 90805 5.375284 5.297322 5.276940
9 95823 4.734783 4.902095 4.838262
10 95823 4.733003 4.896670 4.838262
11 95823 4.732592 4.579438 4.765748
12 94544 4.698490 4.909205 4.838262
13 94544 4.698010 4.897126 4.838262
14 94544 4.697839 4.789904 4.838262
15 94544 4.697600 4.706985 4.838262
16 94544 4.697600 4.706985 4.838262
17 94544 4.695478 4.670329 4.778698
18 90025 5.231136 5.266994 5.216369
In [34]:
agg_df = predictions_df.groupby('Zip Code')['Gradient Boosting'].agg([("Highest", "max"), ("Lowest", "min")]).reset_index()
agg_df.columns = ['Zip Code', 'Highest', 'Lowest']
print(agg_df)

  Zip Code   Highest    Lowest
0    90025  5.216369  5.216369
1    90805  5.304014  5.276940
2    94110  4.909164  4.909164
3    94544  4.838262  4.778698
4    95823  4.838262  4.765748

Top five zip code and prices¶

In [35]:
print("Top five zip code and prices")
print(agg_df)

Top five zip code and prices
  Zip Code   Highest    Lowest
0    90025  5.216369  5.216369
1    90805  5.304014  5.276940
2    94110  4.909164  4.909164
3    94544  4.838262  4.778698
4    95823  4.838262  4.765748