Blight Violations in the City of Detroit¶
Run in Google ColabObjective: Train a Decision Tree classifier to predict whether a given blight ticket will be paid on time.
Import libraries¶
In [1]:
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import ConfusionMatrixDisplay, classification_report, roc_curve, roc_auc_score
import folium
import seaborn as sns
sns.set_style('whitegrid')
Load the dataset¶
In [2]:
data_url = 'https://raw.githubusercontent.com/LuisAngelMendozaVelasco/Applied_Data_Science_with_Python_Specialization/main/Applied_Machine_Learning_in_Python/Week4/Labs/data/train.csv.gz'
df_data = pd.read_csv(data_url, encoding="ISO-8859-1", low_memory=False)
df_data.head()
Out[2]:
| ticket_id | agency_name | inspector_name | violator_name | violation_street_number | violation_street_name | violation_zip_code | mailing_address_str_number | mailing_address_str_name | city | ... | clean_up_cost | judgment_amount | payment_amount | balance_due | payment_date | payment_status | collection_status | grafitti_status | compliance_detail | compliance | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 22056 | Buildings, Safety Engineering & Env Department | Sims, Martinzie | INVESTMENT INC., MIDWEST MORTGAGE | 2900.0 | TYLER | NaN | 3.0 | S. WICKER | CHICAGO | ... | 0.0 | 305.0 | 0.0 | 305.0 | NaN | NO PAYMENT APPLIED | NaN | NaN | non-compliant by no payment | 0.0 |
| 1 | 27586 | Buildings, Safety Engineering & Env Department | Williams, Darrin | Michigan, Covenant House | 4311.0 | CENTRAL | NaN | 2959.0 | Martin Luther King | Detroit | ... | 0.0 | 855.0 | 780.0 | 75.0 | 2005-06-02 00:00:00 | PAID IN FULL | NaN | NaN | compliant by late payment within 1 month | 1.0 |
| 2 | 22062 | Buildings, Safety Engineering & Env Department | Sims, Martinzie | SANDERS, DERRON | 1449.0 | LONGFELLOW | NaN | 23658.0 | P.O. BOX | DETROIT | ... | 0.0 | 0.0 | 0.0 | 0.0 | NaN | NO PAYMENT APPLIED | NaN | NaN | not responsible by disposition | NaN |
| 3 | 22084 | Buildings, Safety Engineering & Env Department | Sims, Martinzie | MOROSI, MIKE | 1441.0 | LONGFELLOW | NaN | 5.0 | ST. CLAIR | DETROIT | ... | 0.0 | 0.0 | 0.0 | 0.0 | NaN | NO PAYMENT APPLIED | NaN | NaN | not responsible by disposition | NaN |
| 4 | 22093 | Buildings, Safety Engineering & Env Department | Sims, Martinzie | NATHANIEL, NEAL | 2449.0 | CHURCHILL | NaN | 7449.0 | CHURCHILL | DETROIT | ... | 0.0 | 0.0 | 0.0 | 0.0 | NaN | NO PAYMENT APPLIED | NaN | NaN | not responsible by disposition | NaN |
5 rows × 34 columns
In [3]:
addresses_url = 'https://raw.githubusercontent.com/LuisAngelMendozaVelasco/Applied_Data_Science_with_Python_Specialization/main/Applied_Machine_Learning_in_Python/Week4/Labs/data/addresses.csv.gz'
df_addresses = pd.read_csv(addresses_url)
df_addresses.head()
Out[3]:
| ticket_id | address | |
|---|---|---|
| 0 | 22056 | 2900 tyler, Detroit MI |
| 1 | 27586 | 4311 central, Detroit MI |
| 2 | 22062 | 1449 longfellow, Detroit MI |
| 3 | 22084 | 1441 longfellow, Detroit MI |
| 4 | 22093 | 2449 churchill, Detroit MI |
In [4]:
latlons_url = 'https://raw.githubusercontent.com/LuisAngelMendozaVelasco/Applied_Data_Science_with_Python_Specialization/main/Applied_Machine_Learning_in_Python/Week4/Labs/data/latlons.csv'
df_latlons = pd.read_csv(latlons_url)
df_latlons.head()
Out[4]:
| address | lat | lon | |
|---|---|---|---|
| 0 | 4300 rosa parks blvd, Detroit MI 48208 | 42.346169 | -83.079962 |
| 1 | 14512 sussex, Detroit MI | 42.394657 | -83.194265 |
| 2 | 3456 garland, Detroit MI | 42.373779 | -82.986228 |
| 3 | 5787 wayburn, Detroit MI | 42.403342 | -82.957805 |
| 4 | 5766 haverhill, Detroit MI | 42.407255 | -82.946295 |
Understand the dataset¶
Blight violations are issued by the city to individuals who allow their properties to remain in a deteriorated condition. Every year, the city of Detroit issues millions of dollars in fines to residents and every year, many of these fines remain unpaid. Enforcing unpaid blight fines is a costly and tedious process.
Each row of the dataset corresponds to a single blight ticket, and includes information about when, why, and to whom each ticket was issued. The target variable is compliance, which is True if the ticket was paid early, on time, or within one month of the hearing data, False if the ticket was paid after the hearing date or not at all, and Null if the violator was found not responsible.
File descriptions:
- train.csv -> the dataset (all tickets issued 2004-2011)
- addresses.csv & latlons.csv -> mapping from ticket id to addresses, and from addresses to lat/lon coordinates.
Data fields:
- ticket_id -> unique identifier for tickets
- agency_name -> Agency that issued the ticket
- inspector_name -> Name of inspector that issued the ticket
- violator_name -> Name of the person/organization that the ticket was issued to
- violation_street_number, violation_street_name, violation_zip_code -> Address where the violation occurred
- mailing_address_str_number, mailing_address_str_name, city, state, zip_code, non_us_str_code, country -> Mailing address of the violator
- ticket_issued_date -> Date and time the ticket was issued
- hearing_date -> Date and time the violator's hearing was scheduled
- violation_code, violation_description -> Type of violation
- disposition -> Judgment and judgement type
- fine_amount -> Violation fine amount, excluding fees
- admin_fee -> $20 fee assigned to responsible judgments
- state_fee -> $10 fee assigned to responsible judgments
- late_fee -> 10% fee assigned to responsible judgments
- discount_amount -> discount applied, if any
- clean_up_cost -> DPW clean-up or graffiti removal cost
- judgment_amount -> Sum of all fines and fees
- grafitti_status -> Flag for graffiti violations
- payment_amount -> Amount paid, if any
- payment_date -> Date payment was made, if it was received
- payment_status -> Current payment status as of Feb 1 2017
- balance_due -> Fines and fees still owed
- collection_status -> Flag for payments in collections
- compliance [target variable for prediction]
- Null = Not responsible
- 0 = Responsible, non-compliant
- 1 = Responsible, compliant
- compliance_detail -> More information on why each ticket was marked compliant or non-compliant
Note: All tickets where the violators were found not responsible are not considered for the prediction task.
In [5]:
df_data.info()
<class 'pandas.core.frame.DataFrame'> RangeIndex: 250306 entries, 0 to 250305 Data columns (total 34 columns): # Column Non-Null Count Dtype --- ------ -------------- ----- 0 ticket_id 250306 non-null int64 1 agency_name 250306 non-null object 2 inspector_name 250306 non-null object 3 violator_name 250272 non-null object 4 violation_street_number 250306 non-null float64 5 violation_street_name 250306 non-null object 6 violation_zip_code 0 non-null float64 7 mailing_address_str_number 246704 non-null float64 8 mailing_address_str_name 250302 non-null object 9 city 250306 non-null object 10 state 250213 non-null object 11 zip_code 250305 non-null object 12 non_us_str_code 3 non-null object 13 country 250306 non-null object 14 ticket_issued_date 250306 non-null object 15 hearing_date 237815 non-null object 16 violation_code 250306 non-null object 17 violation_description 250306 non-null object 18 disposition 250306 non-null object 19 fine_amount 250305 non-null float64 20 admin_fee 250306 non-null float64 21 state_fee 250306 non-null float64 22 late_fee 250306 non-null float64 23 discount_amount 250306 non-null float64 24 clean_up_cost 250306 non-null float64 25 judgment_amount 250306 non-null float64 26 payment_amount 250306 non-null float64 27 balance_due 250306 non-null float64 28 payment_date 41113 non-null object 29 payment_status 250306 non-null object 30 collection_status 36897 non-null object 31 grafitti_status 1 non-null object 32 compliance_detail 250306 non-null object 33 compliance 159880 non-null float64 dtypes: float64(13), int64(1), object(20) memory usage: 64.9+ MB
In [6]:
df_addresses.info()
<class 'pandas.core.frame.DataFrame'> RangeIndex: 311307 entries, 0 to 311306 Data columns (total 2 columns): # Column Non-Null Count Dtype --- ------ -------------- ----- 0 ticket_id 311307 non-null int64 1 address 311307 non-null object dtypes: int64(1), object(1) memory usage: 4.8+ MB
In [7]:
df_latlons.info()
<class 'pandas.core.frame.DataFrame'> RangeIndex: 121769 entries, 0 to 121768 Data columns (total 3 columns): # Column Non-Null Count Dtype --- ------ -------------- ----- 0 address 121769 non-null object 1 lat 121762 non-null float64 2 lon 121762 non-null float64 dtypes: float64(2), object(1) memory usage: 2.8+ MB
Preprocess the dataset¶
In [8]:
df_merged = df_data.merge(df_addresses.merge(df_latlons, on="address"), on="ticket_id")
irrelevant_features = ['ticket_id', 'inspector_name', 'violator_name', 'violation_street_number', 'violation_street_name', 'violation_zip_code',
'mailing_address_str_number', 'mailing_address_str_name', 'city', 'state', 'zip_code', 'non_us_str_code', 'country',
'ticket_issued_date', 'hearing_date', 'violation_description', 'violation_code', 'admin_fee', 'state_fee', 'clean_up_cost',
'grafitti_status', 'payment_amount', 'balance_due', 'payment_date', 'collection_status', 'compliance_detail', 'address']
df_reduced = df_merged[df_merged['country'] == "USA"].drop(irrelevant_features, axis=1).dropna()
df_reduced = pd.get_dummies(df_reduced, drop_first=True)
df_reduced.head()
Out[8]:
| fine_amount | late_fee | discount_amount | judgment_amount | compliance | lat | lon | agency_name_Department of Public Works | agency_name_Detroit Police Department | agency_name_Health Department | agency_name_Neighborhood City Halls | disposition_Responsible by Admission | disposition_Responsible by Default | disposition_Responsible by Determination | payment_status_PAID IN FULL | payment_status_PARTIAL PAYMENT APPLIED | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 250.0 | 25.0 | 0.0 | 305.0 | 0.0 | 42.390729 | -83.124268 | False | False | False | False | False | True | False | False | False |
| 1 | 750.0 | 75.0 | 0.0 | 855.0 | 1.0 | 42.326937 | -83.135118 | False | False | False | False | False | False | True | True | False |
| 5 | 250.0 | 25.0 | 0.0 | 305.0 | 0.0 | 42.145257 | -83.208233 | False | False | False | False | False | True | False | False | False |
| 6 | 750.0 | 75.0 | 0.0 | 855.0 | 0.0 | 42.433466 | -83.023493 | False | False | False | False | False | True | False | False | False |
| 7 | 100.0 | 10.0 | 0.0 | 140.0 | 0.0 | 42.388641 | -83.037858 | False | False | False | False | False | True | False | False | False |
In [9]:
df_reduced.info()
<class 'pandas.core.frame.DataFrame'> Index: 159867 entries, 0 to 250293 Data columns (total 16 columns): # Column Non-Null Count Dtype --- ------ -------------- ----- 0 fine_amount 159867 non-null float64 1 late_fee 159867 non-null float64 2 discount_amount 159867 non-null float64 3 judgment_amount 159867 non-null float64 4 compliance 159867 non-null float64 5 lat 159867 non-null float64 6 lon 159867 non-null float64 7 agency_name_Department of Public Works 159867 non-null bool 8 agency_name_Detroit Police Department 159867 non-null bool 9 agency_name_Health Department 159867 non-null bool 10 agency_name_Neighborhood City Halls 159867 non-null bool 11 disposition_Responsible by Admission 159867 non-null bool 12 disposition_Responsible by Default 159867 non-null bool 13 disposition_Responsible by Determination 159867 non-null bool 14 payment_status_PAID IN FULL 159867 non-null bool 15 payment_status_PARTIAL PAYMENT APPLIED 159867 non-null bool dtypes: bool(9), float64(7) memory usage: 11.1 MB
In [10]:
print("The dataset was reduced by {:.2f}%".format((1 - df_reduced.size / df_merged.size) * 100))
The dataset was reduced by 72.38%
Visualize the class distribution¶
In [11]:
target_feature = df_reduced.columns[4]
class_names = ["non-compliant", "compliant"]
labels, sizes = np.unique(df_reduced[target_feature], return_counts=True)
fig, ax = plt.subplots(figsize=(6, 6))
ax.pie(sizes, textprops={'color': "w", 'fontsize': '12'}, autopct=lambda pct: "{:.2f}%\n({:d})".format(pct, round(pct/100 * sum(sizes))))
ax.legend([str(i) + " (" + class_names[i] + ")" for i in labels.astype(int)])
ax.set_title(target_feature)
plt.show()
Visualize the location of some samples¶
Red markers indicate compliant tickets, while blue markers indicate non-compliant tickets.
In [12]:
map = folium.Map(location=[42.36, -83.10], zoom_start=12)
feature_group = folium.FeatureGroup()
df_sample = df_reduced.sample(n=1000)
for point, compliance in zip(list(df_sample[["lat", "lon"]].to_numpy()), list(df_sample["compliance"])):
if compliance == 1:
feature_group.add_child(folium.Marker(point, popup='compliant', icon=folium.Icon(color='red')))
else:
feature_group.add_child(folium.Marker(point, popup='non-compliant', icon=folium.Icon(color='blue')))
map.add_child(feature_group)
map
Out[12]:
Make this Notebook Trusted to load map: File -> Trust Notebook
Split the dataset into train and test subsets¶
In [13]:
X = df_reduced.drop("compliance", axis=1)
y = df_reduced["compliance"]
X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=0)
print("X_train shape:", X_train.shape)
print("X_test shape:", X_test.shape)
X_train shape: (119900, 15) X_test shape: (39967, 15)
Train a Random Forest classifier¶
In [14]:
random_forest_classifier = RandomForestClassifier()
random_forest_classifier.fit(X_train, y_train)
Out[14]:
RandomForestClassifier()In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.
RandomForestClassifier()
Evaluate the Random Forest classifier¶
In [15]:
y_pred = random_forest_classifier.predict(X_test)
print(classification_report(y_test, y_pred, digits=4))
ConfusionMatrixDisplay.from_predictions(y_test, y_pred, display_labels=[str(i) + " (" + class_names[i] + ")" for i in labels.astype(int)])
plt.grid(False)
plt.show()
precision recall f1-score support
0.0 0.9689 0.9837 0.9762 37064
1.0 0.7416 0.5963 0.6611 2903
accuracy 0.9556 39967
macro avg 0.8553 0.7900 0.8187 39967
weighted avg 0.9524 0.9556 0.9533 39967
A ROC curve is a graphical plot used to illustrate the diagnostic ability of a binary classifier model at various threshold settings. It plots the true positive rate (TPR) against the false positive rate (FPR) at each threshold value. The curve helps in understanding how well the model can distinguish between the positive and negative classes across all possible classification thresholds.
In the context of binary classification, the ROC curve can be used to evaluate the performance of a model by measuring the area under the curve (AUC), which summarizes the classifier's ability to discriminate between positive and negative classes. A higher AUC indicates better performance, with a perfect model having an AUC of 1 and a random model having an AUC of 0.5.
However, it's important to note that the ROC curve and its AUC do not capture certain aspects of performance, such as precision and recall values, and sometimes the inclusion of areas with low sensitivity and specificity can distort the overall performance assessment.
In [16]:
y_proba = random_forest_classifier.predict_proba(X_test)[:, -1]
fpr, tpr, _ = roc_curve(y_test, y_proba)
auc = roc_auc_score(y_test, y_proba)
plt.figure()
plt.plot(fpr, tpr)
plt.fill_between(fpr, tpr, alpha=0.25)
plt.title('ROC curve')
plt.xlabel('FPR')
plt.ylabel('TPR')
plt.text(0.5, 0.5, f'AUC = {auc:.4f}', fontsize=12, bbox=dict(facecolor='white', alpha=0.5))
plt.show()
