Introduction

The data, collected and provided by the City of Austin, Texas, spans from October 1st, 2013, to the present day (updated daily). These datasets comprehensively track the journey of animals from intake to eventual adoption, including those still awaiting adoption. The analysis explores animal adoption trends, rates, influencing factors, and variations in meeting the 90% save rate benchmark on a monthly basis. It assesses adoption rates based on various factors such as age, health conditions, and the impact of external factors, such as the onset of COVID-19, on adoption rates.

This page showcases the results of hypotheses and research questions derived from this analysis. For a comprehensive view, including fetching data using API endpoints, merging datasets, data cleaning, and visualization code, you can explore my Github repository.

# run this if env already exists
library(reticulate)
use_virtualenv("my_python_env")

Load Libraries

Code

import numpy as np
import pandas as pd
import plotly.express as px
import scipy.stats as stats
from data_viz import *
from hypothesis_functions import *

Load Data

Code

data = pd.read_csv("data/animal_center.csv")
data

	animal_id	name	animal_type	sex_intake	sex_outcome	breed	color	date_of_birth	found_location	intake_datetime	outcome_datetime	intake_type	intake_condition	outcome_type	outcome_subtype	age_upon_intake(years)	age_upon_outcome(years)	duration(days)
0	A006100	Scamp	Dog	Neutered Male	Neutered Male	Spinone Italiano Mix	Yellow/White	2007-07-09	8700 Research in Austin (TX)	2014-03-07 14:26:00	2014-03-08 17:10:00	Public Assist	Normal	Return to Owner	NaN	6.7	6.7	1.0
1	A006100	Scamp	Dog	Neutered Male	Neutered Male	Spinone Italiano Mix	Yellow/White	2007-07-09	8700 Research Blvd in Austin (TX)	2014-12-19 10:21:00	2014-12-20 16:35:00	Public Assist	Normal	Return to Owner	NaN	7.4	7.4	1.0
2	A006100	Scamp	Dog	Neutered Male	Neutered Male	Spinone Italiano Mix	Yellow/White	2007-07-09	Colony Creek And Hunters Trace in Austin (TX)	2017-12-07 14:07:00	2017-12-07 00:00:00	Stray	Normal	Return to Owner	NaN	10.4	10.4	-1.0
3	A047759	Oreo	Dog	Neutered Male	Neutered Male	Dachshund	Tricolor	2004-04-02	Austin (TX)	2014-04-02 15:55:00	2014-04-07 15:12:00	Owner Surrender	Normal	Transfer	Partner	10.0	10.0	4.0
4	A134067	Bandit	Dog	Neutered Male	Neutered Male	Shetland Sheepdog	Brown/White	1997-10-16	12034 Research Blvd in Austin (TX)	2013-11-16 09:02:00	2013-11-16 11:54:00	Public Assist	Injured	Return to Owner	NaN	16.1	16.1	0.0
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
157783	A894540	A894540	Dog	Intact Male	NaN	Blue Lacy Mix	Blue/White	NaN	Rutlan And North Lamar in Austin (TX)	2023-12-11 11:24:00	NaN	Stray	Normal	NaN	NaN	NaN	NaN	NaN
157784	A894541	A894541	Cat	Intact Female	NaN	Domestic Shorthair Mix	Black	NaN	2439 Town Lake Circle in Austin (TX)	2023-12-11 12:12:00	NaN	Stray	Normal	NaN	NaN	NaN	NaN	NaN
157785	A894547	A894547	Dog	Intact Female	NaN	Boxer Mix	Brown/White	NaN	6200 Stiles Cove in Austin (TX)	2023-12-11 12:59:00	NaN	Stray	Normal	NaN	NaN	NaN	NaN	NaN
157786	A894551	A894551	Cat	Intact Female	NaN	Domestic Shorthair Mix	Tortie	NaN	3002 Cheviot Lane in Austin (TX)	2023-12-11 13:20:00	NaN	Stray	Normal	NaN	NaN	NaN	NaN	NaN
157787	A894558	NaN	Dog	Intact Female	NaN	Staffordshire Mix	White/Black	NaN	5301 Decker Lane in Austin (TX)	2023-12-11 15:27:00	NaN	Stray	Normal	NaN	NaN	NaN	NaN	NaN

157788 rows × 18 columns

Exploratory Data Analysis

Code

data.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 157788 entries, 0 to 157787
Data columns (total 18 columns):
 #   Column                   Non-Null Count   Dtype  
---  ------                   --------------   -----  
 0   animal_id                157788 non-null  object 
 1   name                     112350 non-null  object 
 2   animal_type              157788 non-null  object 
 3   sex_intake               157786 non-null  object 
 4   sex_outcome              156812 non-null  object 
 5   breed                    157788 non-null  object 
 6   color                    157788 non-null  object 
 7   date_of_birth            156814 non-null  object 
 8   found_location           157788 non-null  object 
 9   intake_datetime          157788 non-null  object 
 10  outcome_datetime         156814 non-null  object 
 11  intake_type              157788 non-null  object 
 12  intake_condition         157788 non-null  object 
 13  outcome_type             156783 non-null  object 
 14  outcome_subtype          72131 non-null   object 
 15  age_upon_intake(years)   156814 non-null  float64
 16  age_upon_outcome(years)  156814 non-null  float64
 17  duration(days)           156814 non-null  float64
dtypes: float64(3), object(15)
memory usage: 21.7+ MB

Animal Type

Code

animal_type_freq = data.animal_type.value_counts().reset_index()
animal_type_freq
freq_bar_plot(animal_type_freq, "animal_type", "count")

Sex Intake

Code

sex_type_intake_freq = data.sex_intake.value_counts().reset_index()
freq_bar_plot(sex_type_intake_freq, "sex_intake", "count")

Sex Outcome

Code

sex_type_outcome_freq = data.sex_outcome.value_counts().reset_index()
freq_bar_plot(sex_type_outcome_freq, "sex_outcome", "count")

Code

sex_type_freq = pd.merge(sex_type_intake_freq, sex_type_outcome_freq,
                        left_on = "sex_intake",
                        right_on = "sex_outcome",
                        suffixes = ["_intake", "_outcome"])
sex_type_freq.drop(columns=['sex_outcome'], inplace=True)
sex_type_freq.rename(columns = {"sex_intake":"sex"}, inplace=True)
# sex_type_freq

fig = px.bar(sex_type_freq,
            x="sex",
            y=["count_intake", "count_outcome"],
            barmode="group",
            labels={"sex": "Sex", "value": "Count", "variable": "Category"},
            title="Count of Intake and Outcome by Sex")

fig.update_traces(name="Intake",
                  text=sex_type_freq["count_intake"],
                  textposition="outside",
                  selector=dict(name="count_intake"),
                  marker=dict(color="#13294B"))
fig.update_traces(name="Outcome",
                  text=sex_type_freq["count_outcome"],
                  textposition="outside",
                  selector=dict(name="count_outcome"),
                  marker=dict(color="#E84A27"))
fig.update_layout(title_x=0.5,
                 height=600,
                 width=1000)
fig.show()

Intake Condition

Code

intake_cond_freq = data.intake_condition.value_counts().reset_index()
freq_bar_plot(intake_cond_freq, "intake_condition", "count")

Outcome Type

Code

outcome_type_freq = data.outcome_type.value_counts().reset_index()
freq_bar_plot(outcome_type_freq, "outcome_type", "count")

Number of Intake Overtime

Code

fig = px.histogram(data.intake_datetime,
                   nbins=600,
                   title="Animal Intake Distribution",
                   labels={"value":"date",
                           "count":"Counts"})
fig.update_traces(marker_color="#13294B",
                  hovertemplate='Date: %{x|%Y-%m}<br>Counts: %{y}')
fig.update_layout(title_x=0.5,
                  showlegend=False)
fig.show()

Number of Outcome Overtime

Code

fig = px.histogram(data.outcome_datetime,
                   nbins=600,
                   title="Animal Outcome Distribution",
                   labels={"value":"date",
                           "count":"Counts"})
fig.update_traces(marker_color="#13294B",
                  hovertemplate='Date: %{x|%Y-%m}<br>Counts: %{y}')
fig.update_layout(title_x=0.5,
                  showlegend=False)
fig.show()

Hypotheses

Hypothesis1:

Is there a correlation between the neuter/spay rate of animals and their age?

\(H0\): There is no significant relationship between neuter/spay rate and age in animals.
\(H1\): There is a significant relationship between neuter/spay rate and age in animals.

While there isn’t a specific age limit for the spaying or neutering procedure, the recommended timeframe for dogs or cats is typically between six to nine months. Concerns may arise regarding the risks associated with surgery in older animals. Additionally, if an animal hasn’t exhibited behavioral issues related to mating behaviors or health problems linked to remaining intact, some owners might perceive less urgency in proceeding with the surgery.

Code

data.outcome_type.value_counts()

outcome_type
Adoption           73997
Transfer           44944
Return to Owner    24440
Euthanasia          9967
Died                1470
Rto-Adopt           1082
Disposal             768
Missing               84
Relocate              26
Stolen                 5
Name: count, dtype: int64

Code

alive_outcome_type = ["Euthanasia", "Died", "Disposal", "Missing", "Stolen"]
sex_change = data[(data["animal_type"].isin(["Dog", "Cat"])) & 
                  (data["sex_intake"].isin(["Intact Female", "Intact Male"])) &
                  (~data["outcome_type"].isin(alive_outcome_type)) &
                  (~data["age_upon_outcome(years)"].isna())]\
    [["animal_id", "animal_type", "sex_intake", "sex_outcome", "age_upon_outcome(years)", "outcome_datetime", "duration(days)", "outcome_type"]].reset_index(drop=True)

# add neutered_spayed_or_not column
sex_change["neutered_spayed_or_not"] = np.where(sex_change["sex_outcome"].isin(["Neutered Male", "Spayed Female"]), 1, 0)
sex_change

	animal_id	animal_type	sex_intake	sex_outcome	age_upon_outcome(years)	outcome_datetime	duration(days)	outcome_type	neutered_spayed_or_not
0	A163459	Dog	Intact Female	Intact Female	15.1	2014-11-14 19:28:00	0.0	Return to Owner	0
1	A191351	Cat	Intact Female	Intact Female	16.2	2015-11-17 13:29:00	3.0	Return to Owner	0
2	A212672	Dog	Intact Female	Intact Female	13.8	2013-12-06 14:34:00	10.0	Return to Owner	0
3	A256412	Dog	Intact Male	Intact Male	16.6	2013-10-06 14:26:00	0.0	Return to Owner	0
4	A309829	Dog	Intact Male	Intact Male	13.0	2014-11-12 15:34:00	2.0	Return to Owner	0
...	...	...	...	...	...	...	...	...	...
97614	A894297	Dog	Intact Male	Intact Male	2.0	2023-12-06 19:24:00	0.0	Transfer	0
97615	A894324	Dog	Intact Male	Intact Male	0.5	2023-12-08 14:52:00	1.0	Transfer	0
97616	A894377	Cat	Intact Female	Intact Female	1.0	2023-12-08 15:26:00	1.0	Transfer	0
97617	A894389	Dog	Intact Male	Intact Male	10.0	2023-12-09 12:55:00	1.0	Return to Owner	0
97618	A894500	Dog	Intact Male	Intact Male	1.0	2023-12-10 12:43:00	0.0	Transfer	0

97619 rows × 9 columns

Code

neutered_n_spayed_count = sex_change.neutered_spayed_or_not.sum()
animal_count = sex_change.animal_id.count()
neutered_n_spayed_pct = neutered_n_spayed_count/animal_count

print(f"{round(neutered_n_spayed_pct, 3)*100.0}% of the animals that were intact before have been neutered/spayed upon outcome.")

64.0% of the animals that were intact before have been neutered/spayed upon outcome.

Independent Samples T-Test

The Independent Samples T-Test is employed to compare the means of ages between two independent groups (neutered/spayed or not). This test helps define whether a significant difference exists in age between these two distinct groups.

Code

neutered_spayed = sex_change[sex_change["neutered_spayed_or_not"] == 1]["age_upon_outcome(years)"]
not_neutered_spayed = sex_change[sex_change["neutered_spayed_or_not"] == 0]["age_upon_outcome(years)"]

Code

# Perform Independent Samples T-Test
t_statistic, p_value = stats.ttest_ind(neutered_spayed, not_neutered_spayed)

# Display results
print(f"T-Statistic: {t_statistic}")
print(f"P-Value: {p_value}")

T-Statistic: -27.800901329031614
P-Value: 1.9453487884732793e-169

Interpretation: With such a small p-value (close to zero), we reject the null hypothesis. Therefore, there is strong evidence to suggest that there is a significant difference in age between animals that are neutered/spayed and those that are not neutered/spayed. The negative T-Statistic value indicates that, on average, the age of animals that are neutered/spayed is significantly lower than the age of animals that are not neutered/spayed.

Chi-Squared Test

The Chi-Squared Test is utilized to ascertain whether an association exists between two categorical variables (age group and neutered/spayed status). This test evaluates the relationship between these categorical variables to determine if they are associated or independent of each other.

Code

sex_change["age_group"] = pd.cut(sex_change["age_upon_outcome(years)"],
                                 bins=[-0.01, 2, 7, float('inf')],
                                 labels=["Young", "Adult", "Senior"])

sex_change

	animal_id	animal_type	sex_intake	sex_outcome	age_upon_outcome(years)	outcome_datetime	duration(days)	outcome_type	neutered_spayed_or_not	age_group
0	A163459	Dog	Intact Female	Intact Female	15.1	2014-11-14 19:28:00	0.0	Return to Owner	0	Senior
1	A191351	Cat	Intact Female	Intact Female	16.2	2015-11-17 13:29:00	3.0	Return to Owner	0	Senior
2	A212672	Dog	Intact Female	Intact Female	13.8	2013-12-06 14:34:00	10.0	Return to Owner	0	Senior
3	A256412	Dog	Intact Male	Intact Male	16.6	2013-10-06 14:26:00	0.0	Return to Owner	0	Senior
4	A309829	Dog	Intact Male	Intact Male	13.0	2014-11-12 15:34:00	2.0	Return to Owner	0	Senior
...	...	...	...	...	...	...	...	...	...	...
97614	A894297	Dog	Intact Male	Intact Male	2.0	2023-12-06 19:24:00	0.0	Transfer	0	Young
97615	A894324	Dog	Intact Male	Intact Male	0.5	2023-12-08 14:52:00	1.0	Transfer	0	Young
97616	A894377	Cat	Intact Female	Intact Female	1.0	2023-12-08 15:26:00	1.0	Transfer	0	Young
97617	A894389	Dog	Intact Male	Intact Male	10.0	2023-12-09 12:55:00	1.0	Return to Owner	0	Senior
97618	A894500	Dog	Intact Male	Intact Male	1.0	2023-12-10 12:43:00	0.0	Transfer	0	Young

97619 rows × 10 columns

Code

neuter_spay_age_contingency_table = pd.crosstab(sex_change["age_group"], sex_change["neutered_spayed_or_not"])
neuter_spay_age_contingency_table

neutered_spayed_or_not	0	1
age_group
Young	29165	52646
Adult	4366	9131
Senior	1573	730

Code

age_groups = ["Young", "Adult", "Senior"]

for age_group in age_groups:
    total = neuter_spay_age_contingency_table.loc[age_group].sum()
    neuter_spay_count = neuter_spay_age_contingency_table.loc[age_group, 1]
    neuter_spay_rate = round(neuter_spay_count/total, 4)*100.0
    print(f"The neuter/spay rate for {age_group.lower()} dogs/cats is {neuter_spay_rate}%")

The neuter/spay rate for young dogs/cats is 64.35%
The neuter/spay rate for adult dogs/cats is 67.65%
The neuter/spay rate for senior dogs/cats is 31.7%

Code

chi2, p, dof, expected = stats.chi2_contingency(neuter_spay_age_contingency_table)
print(f"chi2:    {chi2}")
print(f"p-value: {p}")
print(f"dof:     {dof}")
print(f"expected:\n{expected}")

# set significance level = 0.05
alpha = 0.05
if p < alpha:
    print("\nThere is a significant relationship between age and neutered/spayed status.")
else:
    print("\nThere is no significant relationship between age and neutered/spayed status.")

chi2:    1125.9288581014591
p-value: 3.218510686406676e-245
dof:     2
expected:
[[29421.82073742 52389.17926258]
 [ 4853.94769032  8643.05230968]
 [  828.23157226  1474.76842774]]

There is a significant relationship between age and neutered/spayed status.

Code

fig = px.histogram(sex_change,
                   x="age_group",
                   color="neutered_spayed_or_not",
                   barmode="group",
                   title="Neutered/Spayed Cases by Age Group")

fig.update_layout(title_x=0.5,
                  xaxis=dict(title="Age Group"),
                  yaxis=dict(title="Counts"),
                  legend_title_text="Neutered/Spayed Status",
                  height=600,
                  width=1000)

fig.update_layout(legend=dict(orientation="v",
                              yanchor="bottom",
                              y=1.02,
                              xanchor="right",
                              x=1))

fig.show()

Hypothesis2:

Has there been a significant change in the adoption rate of animals before and after the onset of COVID-19?

\(H0\): There is no significant difference in the adoption rate of animals before and after the onset of COVID.
\(H1\): There is a significant increase in the adoption rate of animals after the onset of COVID compared to before.

With more individuals spending extended periods at home due to lockdowns and remote work scenarios, there might have been an increased desire for companionship, potentially leading to a higher demand for pets

Code

adoption_research = data[["animal_id", "outcome_datetime", "outcome_type"]].reset_index(drop=True)
adoption_research = adoption_research.dropna(subset=["outcome_datetime"])
adoption_research["adoption_status"] = np.where(adoption_research["outcome_type"] == "Adoption", "Adopted", "Not Adopted")
datetime_threshold = "2020-01-01"
adoption_research["before_or_after_covid"] = np.where(adoption_research["outcome_datetime"] < datetime_threshold, "pre-covid", "post-covid")
adoption_research

	animal_id	outcome_datetime	outcome_type	adoption_status	before_or_after_covid
0	A006100	2014-03-08 17:10:00	Return to Owner	Not Adopted	pre-covid
1	A006100	2014-12-20 16:35:00	Return to Owner	Not Adopted	pre-covid
2	A006100	2017-12-07 00:00:00	Return to Owner	Not Adopted	pre-covid
3	A047759	2014-04-07 15:12:00	Transfer	Not Adopted	pre-covid
4	A134067	2013-11-16 11:54:00	Return to Owner	Not Adopted	pre-covid
...	...	...	...	...	...
157763	A894445	2023-12-10 15:45:00	Adoption	Adopted	post-covid
157768	A894472	2023-12-11 07:46:00	Transfer	Not Adopted	post-covid
157773	A894500	2023-12-10 12:43:00	Transfer	Not Adopted	post-covid
157774	A894504	2023-12-11 12:39:00	Return to Owner	Not Adopted	post-covid
157776	A894514	2023-12-11 08:40:00	Euthanasia	Not Adopted	post-covid

156814 rows × 5 columns

Code

adoption_contingency_table = pd.crosstab(adoption_research["before_or_after_covid"], adoption_research["adoption_status"])
adoption_contingency_table

adoption_status	Adopted	Not Adopted
before_or_after_covid
post-covid	24716	19476
pre-covid	49281	63341

Code

covid_list = ["post-covid", "pre-covid"]

for covid in covid_list:
    total = adoption_contingency_table.loc[covid].sum()
    adoption_count = adoption_contingency_table.loc[covid, "Adopted"]
    adoption_rate = round(adoption_count/total, 4)*100.0
    print(f"The adoption rate {covid} is {adoption_rate}%")

The adoption rate post-covid is 55.93%
The adoption rate pre-covid is 43.76%

Code

chi2, p, dof, expected = stats.chi2_contingency(adoption_contingency_table)
print(f"chi2:    {chi2}")
print(f"p-value: {p}")
print(f"dof:     {dof}")
print(f"expected:\n{expected}")

# set significance level = 0.05
alpha = 0.05
if p < alpha:
    print("\nThere is a significant increase in the adoption rate post-COVID compared to pre-COVID.")
else:
    print("\nThere is no significant increase in the adoption rate post-COVID compared to pre-COVID.")

chi2:    1886.008048757299
p-value: 0.0
dof:     1
expected:
[[20853.21096331 23338.78903669]
 [53143.78903669 59478.21096331]]

There is a significant increase in the adoption rate post-COVID compared to pre-COVID.

Code

fig = px.histogram(adoption_research,
             x="before_or_after_covid",
             color="adoption_status",
             barmode = "group",
             title="Adoption Cases Before and After Covid")

fig.update_layout(title_x=0.5,
                  xaxis=dict(title="Adoption Status"),
                  yaxis=dict(title="Counts"),
                  legend_title_text="Adoption Status",
                  height=600,
                  width=1000)

fig.update_layout(legend=dict(
    orientation="v",
    yanchor="bottom",
    y=1.02,
    xanchor="right",
    x=1))

fig.show()

Hypothesis3:

Does intake condition influence the probability of adoption for animals?

\(H0\): There is no significant relationship between the intake condition of animals at the shelter and their likelihood of adoption.
\(H1\): There is a significant relationship between the intake condition of animals at the shelter and their likelihood of adoption.

Typically, healthy animals often experience a higher likelihood of swift adoption, whereas those requiring extensive medical attention, facing chronic health issues, or exhibiting challenging behaviors might encounter prolonged waiting periods before finding homes.

Code

abnormal_cond = ["Injured", "Sick", "Nursing", "Neonatal", "Aged", "Medical",
                 "Feral", "Pregnant", "Behavior", "Med Attn", "Med Urgent",
                 "Neurologic", "Space", "Parvo", "Agonal", "Panleuk", "Congenital"]

condition_research = data[~data["intake_condition"].isin(["Other", "Unknown"])]\
                    [["animal_id", "intake_condition", "outcome_type"]].reset_index(drop=True)

# add intake_cond_binary column
condition_research["intake_cond_binary"] = np.where(condition_research["intake_condition"].isin(abnormal_cond), "Non-Normal", "Normal")

# add adoption_status column
condition_research["adoption_status"] = np.where(condition_research["outcome_type"]=="Adoption", "Adopted", "Not Adopted")
condition_research

	animal_id	intake_condition	outcome_type	intake_cond_binary	adoption_status
0	A006100	Normal	Return to Owner	Normal	Not Adopted
1	A006100	Normal	Return to Owner	Normal	Not Adopted
2	A006100	Normal	Return to Owner	Normal	Not Adopted
3	A047759	Normal	Transfer	Normal	Not Adopted
4	A134067	Injured	Return to Owner	Non-Normal	Not Adopted
...	...	...	...	...	...
157427	A894540	Normal	NaN	Normal	Not Adopted
157428	A894541	Normal	NaN	Normal	Not Adopted
157429	A894547	Normal	NaN	Normal	Not Adopted
157430	A894551	Normal	NaN	Normal	Not Adopted
157431	A894558	Normal	NaN	Normal	Not Adopted

157432 rows × 5 columns

Code

condition_adoption_contingency_table = pd.crosstab(condition_research["intake_cond_binary"], condition_research["adoption_status"])
condition_adoption_contingency_table

adoption_status	Adopted	Not Adopted
intake_cond_binary
Non-Normal	5987	16629
Normal	67872	66944

Code

condition_list = ["Non-Normal", "Normal"]

for condition in condition_list:
    total = condition_adoption_contingency_table.loc[condition].sum()
    adoption_cnt = condition_adoption_contingency_table.loc[condition, "Adopted"]
    adoption_rate = round(adoption_cnt/total, 3)*100.0
    print(f"{adoption_rate}% of animal whose intake condition were {condition.lower()} have been adopted.")

26.5% of animal whose intake condition were non-normal have been adopted.
50.3% of animal whose intake condition were normal have been adopted.

Code

chi2, p, dof, expected = stats.chi2_contingency(condition_adoption_contingency_table)
print(f"chi2:    {chi2}")
print(f"p-value: {p}")
print(f"dof:     {dof}")
print(f"expected:\n{expected}")

# set significance level = 0.05
alpha = 0.05
if p < alpha:
    print("\nThere is a significant relationship between the intake condition of animals at the shelter and their likelihood of adoption.")
else:
    print("\nThere is no significant relationship between the intake condition of animals at the shelter and their likelihood of adoption.")

chi2:    4430.532316853005
p-value: 0.0
dof:     1
expected:
[[10610.26439352 12005.73560648]
 [63248.73560648 71567.26439352]]

There is a significant relationship between the intake condition of animals at the shelter and their likelihood of adoption.

Code

fig = px.histogram(condition_research,
             x="intake_cond_binary",
             color="adoption_status",
             barmode = "group",
             title="Adoption Status by Intake Condition")

fig.update_layout(title_x=0.5,
                  xaxis=dict(title="Adoption Status"),
                  yaxis=dict(title="Counts"),
                  legend_title_text="Adoption Status",
                  height=600,
                  width=1000)

fig.update_layout(legend=dict(
    orientation="v",
    yanchor="bottom",
    y=1.02,
    xanchor="right",
    x=1))

fig.show()

Hypothesis4:

Does age influence the probability of adoption for animals?

\(H0\): There is no significant relationship between animals’ age and their likelihood of adoption.
\(H1\): There is a significant relationship between animals’ age and their likelihood of adoption.

An animal’s age could potentially impact its likelihood of adoption. Generally, most people tend to prefer younger animals as they are often perceived as more trainable. Younger animals haven’t developed habits, making them easier to train. Additionally, they offer longer companionship and possess more energy.

Code

age_adoption_research = data[data["age_upon_outcome(years)"] >= 0][["animal_id", "outcome_type", "age_upon_outcome(years)"]].reset_index(drop=True)
age_adoption_research.loc[:, "age_group"] = pd.cut(age_adoption_research["age_upon_outcome(years)"],
                                                    bins=[-0.01, 2, 7, float('inf')],
                                                    labels=["Young", "Adult", "Senior"])
age_adoption_research["adoption_status"] = np.where(age_adoption_research["outcome_type"] == "Adoption", "Adopted", "Not Adopted")
age_adoption_research

	animal_id	outcome_type	age_upon_outcome(years)	age_group	adoption_status
0	A006100	Return to Owner	6.7	Adult	Not Adopted
1	A006100	Return to Owner	7.4	Senior	Not Adopted
2	A006100	Return to Owner	10.4	Senior	Not Adopted
3	A047759	Transfer	10.0	Senior	Not Adopted
4	A134067	Return to Owner	16.1	Senior	Not Adopted
...	...	...	...	...	...
156784	A894445	Adoption	1.5	Young	Adopted
156785	A894472	Transfer	2.0	Young	Not Adopted
156786	A894500	Transfer	1.0	Young	Not Adopted
156787	A894504	Return to Owner	17.0	Senior	Not Adopted
156788	A894514	Euthanasia	2.0	Young	Not Adopted

156789 rows × 5 columns

Independent Samples T-Test

The Independent Samples T-Test is employed to compare the means of ages between two independent groups (being adopted or not). This test helps define whether a significant difference exists in age between these two distinct groups.

Code

adopted = age_adoption_research[age_adoption_research["adoption_status"] == "Adopted"]["age_upon_outcome(years)"]
not_adopted = age_adoption_research[age_adoption_research["adoption_status"] == "Not Adopted"]["age_upon_outcome(years)"]

Code

# Perform Independent Samples T-Test
t_statistic, p_value = stats.ttest_ind(adopted, not_adopted)

# Display results
print(f"T-Statistic: {t_statistic}")
print(f"P-Value: {p_value}")

T-Statistic: -55.12214842454163
P-Value: 0.0

Interpretation: With such a small p-value (equals to zero), we reject the null hypothesis. Therefore, there is strong evidence to suggest that there is a significant difference in age between animals that are adopted and those that are not adopted. The negative T-Statistic value indicates that, on average, the age of animals that are not adopted tends to have a statistically significantly higher mean age compared to the adopted group.

Chi-Squared Test

The Chi-Squared Test is utilized to ascertain whether an association exists between two categorical variables (age group and adoption status). This test evaluates the relationship between these categorical variables to determine if they are associated or independent of each other.

Code

age_adoption_contingency_table = pd.crosstab(age_adoption_research["age_group"], age_adoption_research["adoption_status"])
age_adoption_contingency_table

adoption_status	Adopted	Not Adopted
age_group
Young	55851	56359
Adult	14234	18409
Senior	3903	8033

Code

age_groups = ["Young", "Adult", "Senior"]

for age_group in age_groups:
    total = age_adoption_contingency_table.loc[age_group].sum()
    adoption_count = age_adoption_contingency_table.loc[age_group, "Adopted"]
    adoption_count_rate = round(adoption_count/total, 4)*100.0
    print(f"The adoption rate for {age_group.lower()} dogs/cats is {adoption_count_rate}%")

The adoption rate for young dogs/cats is 49.769999999999996%
The adoption rate for adult dogs/cats is 43.61%
The adoption rate for senior dogs/cats is 32.7%

Code

chi2, p, dof, expected = stats.chi2_contingency(age_adoption_contingency_table)
print(f"chi2:    {chi2}")
print(f"p-value: {p}")
print(f"dof:     {dof}")
print(f"expected:\n{expected}")

# set significance level = 0.05
alpha = 0.05
if p < alpha:
    print("\nThere is a significant relationship between age and adoption status.")
else:
    print("\nThere is no significant relationship between age and adoption status.")

chi2:    1474.593471046482
p-value: 0.0
dof:     2
expected:
[[52951.37720121 59258.62279879]
 [15404.07990356 17238.92009644]
 [ 5632.54289523  6303.45710477]]

There is a significant relationship between age and adoption status.

Code

fig = px.histogram(age_adoption_research,
             x="age_group",
             color="adoption_status",
             barmode = "group",
             title="Adoption Status by Intake Condition")

fig.update_layout(title_x=0.5,
                  xaxis=dict(title="Adoption Status"),
                  yaxis=dict(title="Counts"),
                  legend_title_text="Adoption Status",
                  height=600,
                  width=1000)

fig.update_layout(legend=dict(
    orientation="v",
    yanchor="bottom",
    y=1.02,
    xanchor="right",
    x=1))

fig.show()

Research Questions

Research Question1:

Does the city of Austin maintain a save rate of 90% or higher for animals, thereby meeting the criteria to be classified as a no-kill city?

Austin stands as one of the largest no-kill cities in America. According to the City of Austin, annually, over 90% of animals entering the center are either adopted, transferred to rescues, or returned to their owners. The research question aims to ascertain whether the Austin Animal Center meets this requirement or not.

Code

data.outcome_type.value_counts()

outcome_type
Adoption           73997
Transfer           44944
Return to Owner    24440
Euthanasia          9967
Died                1470
Rto-Adopt           1082
Disposal             768
Missing               84
Relocate              26
Stolen                 5
Name: count, dtype: int64

Code

no_kill_research = data[["animal_id", "intake_datetime", "outcome_datetime", "outcome_type", "duration(days)"]].copy()
create_time_column(no_kill_research, "intake_datetime", "outcome_datetime")

	animal_id	intake_datetime	outcome_datetime	outcome_type	duration(days)	intake_y_m	outcome_y_m	intake_yr	outcome_yr	intake_m	outcome_m
0	A006100	2014-03-07 14:26:00	2014-03-08 17:10:00	Return to Owner	1.0	2014-03	2014-03	2014	2014	March	March
1	A006100	2014-12-19 10:21:00	2014-12-20 16:35:00	Return to Owner	1.0	2014-12	2014-12	2014	2014	December	December
2	A006100	2017-12-07 14:07:00	2017-12-07 00:00:00	Return to Owner	-1.0	2017-12	2017-12	2017	2017	December	December
3	A047759	2014-04-02 15:55:00	2014-04-07 15:12:00	Transfer	4.0	2014-04	2014-04	2014	2014	April	April
4	A134067	2013-11-16 09:02:00	2013-11-16 11:54:00	Return to Owner	0.0	2013-11	2013-11	2013	2013	November	November
...	...	...	...	...	...	...	...	...	...	...	...
157783	A894540	2023-12-11 11:24:00	NaT	NaN	NaN	2023-12	NaT	2023	NaT	December	NaN
157784	A894541	2023-12-11 12:12:00	NaT	NaN	NaN	2023-12	NaT	2023	NaT	December	NaN
157785	A894547	2023-12-11 12:59:00	NaT	NaN	NaN	2023-12	NaT	2023	NaT	December	NaN
157786	A894551	2023-12-11 13:20:00	NaT	NaN	NaN	2023-12	NaT	2023	NaT	December	NaN
157787	A894558	2023-12-11 15:27:00	NaT	NaN	NaN	2023-12	NaT	2023	NaT	December	NaN

157788 rows × 11 columns

Group by year-month

Code

no_kill_by_ym = no_kill_rate(no_kill_research, "intake_y_m", "outcome_y_m")
no_kill_by_ym

	time	total_intake	total_outcome	euthanasia_cases	total_death_cases	save_rate(total death)	save_rate(euthanasia)
0	2013-10	1589.0	1126	147	162	89.804909	90.748899
1	2013-11	1323.0	1213	120	129	90.249433	90.929705
2	2013-12	1270.0	1431	163	171	86.535433	87.165354
3	2014-01	1271.0	1224	114	124	90.243902	91.030685
4	2014-02	1192.0	1115	144	155	86.996644	87.919463
...	...	...	...	...	...	...	...
118	2023-08	965.0	1108	31	56	94.196891	96.787565
119	2023-09	927.0	874	33	46	95.037756	96.440129
120	2023-10	960.0	843	35	45	95.312500	96.354167
121	2023-11	761.0	833	36	56	92.641261	95.269382
122	2023-12	247.0	275	4	6	97.570850	98.380567

123 rows × 7 columns

Code

save_rate_line_plot(no_kill_by_ym, "Overtime")

Group by year

Code

no_kill_by_yr = no_kill_rate(no_kill_research, "intake_yr", "outcome_yr")
no_kill_by_yr

	time	total_intake	total_outcome	euthanasia_cases	total_death_cases	save_rate(total death)	save_rate(euthanasia)
0	2013	4182.0	3770	430	462	88.952654	89.717838
1	2014	18654.0	18559	1844	2080	88.849576	90.114721
2	2015	18709.0	18485	1461	1684	90.998984	92.190924
3	2016	17673.0	17662	1161	1434	91.885928	93.430657
4	2017	17559.0	17643	1083	1327	92.442622	93.832223
5	2018	16975.0	16734	808	1036	93.896907	95.240059
6	2019	19726.0	19769	788	1127	94.286728	96.005272
7	2020	9585.0	9773	698	892	90.693792	92.717788
8	2021	12040.0	11962	597	789	93.446844	95.041528
9	2022	11891.0	11882	607	770	93.524514	94.895299
10	2023	10794.0	10575	490	693	93.579767	95.460441

Code

save_rate_line_plot(no_kill_by_yr, "by Year")

Group by month

Code

no_kill_by_m = no_kill_rate(no_kill_research, "intake_m", "outcome_m")
month_order = ["January", "February", "March", "April",
               "May", "June", "July", "August",
               "September", "October", "November", "December"]
no_kill_by_m["time"] = pd.Categorical(no_kill_by_m["time"],
                                      categories=month_order,
                                      ordered=True)
no_kill_by_m = no_kill_by_m.sort_values("time").reset_index(drop=True)
no_kill_by_m

	time	total_intake	total_outcome	euthanasia_cases	total_death_cases	save_rate(total death)	save_rate(euthanasia)
0	January	10831	11418	555	660	93.906380	94.875819
1	February	9835	9974	679	776	92.109812	93.096085
2	March	11944	11685	1427	1692	85.833891	88.052579
3	April	12479	11213	929	1080	91.345460	92.555493
4	May	16165	13519	913	1140	92.947727	94.351995
5	June	15733	15093	1042	1332	91.533719	93.376978
6	July	14429	15330	820	1096	92.404186	94.317000
7	August	13990	14991	726	982	92.980701	94.810579
8	September	13727	13670	686	881	93.581992	95.002550
9	October	14931	14179	907	1110	92.565803	93.925390
10	November	12591	12971	700	846	93.280915	94.440473
11	December	11133	12771	583	699	93.721369	94.763316

Code

save_rate_line_plot(no_kill_by_m, "by Month")