RPubs link information

• Rpubs link comes here: https://rpubs.com/JohnLin/542718

Introduction

• Physical activity is good for people’s health.
• Sports facilities can assist people in sports activities.
• Good Facility Condition helps people to complete sport activities safely and effectively.
• The purpose of this prsentation is to fugure out the factoes which would affect Facility Condition.
• Find out if the Street Type and Facility Age are associated with the Facility Condition.
• Street Type: beside a street/in a venue/in a park/etc.
• Facility Age: How many years the facility has been used for.

Problem Statement

• Use Chi-square Test of Association to check if there is a association between Street Type and Facility Condition.
• Use Chi-square Test of Association to check if there is a association between Facility Age and Facility Condition.

Data

• We used data provided by the Victorian Government (https://discover.data.vic.gov.au/dataset/sport-and-recreational-facilities )
• The data set includes the coordinates of the facility, ID, name, address, age, condition, etc.
• The data set records a total of 9472 observations
• The survey method: census
• Last Updated:28/07/2016
• Organization: Department of Health and Human Services

Data Cont.

• Filter out the observations with recorded FacilityCondition.
• Select the variables what we want (StreetType, FacilityAge and Facility Condition).
• Clean the Street Type: Integrate similar street types
  \(e.g.\)
  BLVD => BOULEVARD
  ROAD DRIVE, ROAD SOUTH and ROAD WEST => ROAD.
• Choose top 4 types (“AVENUE”, “HIGHWAY”, “ROAD” and “STREET”) to do the analysis.
• Exclude the observations with value “Same as above” because only 5 such observations in the dataframe.
• No missing value in the dataframe.

Data Cont.

• Factorise the three variables.
• StreetType: “AVENUE”, “HIGHWAY”, “ROAD”, “STREET”
• FacilityAge: “0 - 5 Years”, “6 - 10 Years”, “11 - 25 Years”, “26 - 50 Years”, “51 Years +” (ordered)
• FacilityCondition: “Very Poor”, “Poor”, “Average”, “Good”, “Very Good” (ordered)
• the dimension of the dataframe after all preprocessing is 5061 X 3

Descriptive Statistics and Visualisation

• The most street type is ROAD and the second one is STREET
• The number of the other two (AVENUE and HIGHWAY) are much fewer.

Decsriptive Statistics Cont.

• Majority(Over 2000) of facility age is between 26 and 50.
• About 1400 facilities age was constructed from 11 to 25 ago.
• The number of the remainings are all below 500.

Decsriptive Statistics Cont.

• Only a few facilities is recorded as “Very Poor” condition.
• Majority of facility conditions are “Average” or “Good”.

Decsriptive Statistics Cont.

• It seems that there no relation between FacilityCondition and StreetType.

Decsriptive Statistics Cont.

• It is obvious that younger facilities usually have better facility condition.

Hypothesis Testing

• Because no more than 25% of the cells have expected counts below 5, we can use the Chi-square test of association to test the following hypothesis.
• \(H_0:\) There is no association in the population between FacilityCondition and StreetType.
• \(H_A:\) There is an association in the population between FacilityCondition and StreetType.

chi2_con_street <- chisq.test(table(df$FacilityCondition, df$StreetType))
chi2_con_street

## 
##  Pearson's Chi-squared test
## 
## data:  table(df$FacilityCondition, df$StreetType)
## X-squared = 12.446, df = 12, p-value = 0.4105

• Because \(p\)-value = 0.4105 > \(\alpha\) = 0.05, we fail to reject \(H_0:\) There is no association in the population between FacilityCondition and StreetType. As a result, there is no sufficient statistically significant evidence to support \(H_A\).

Hypthesis Testing Cont.

• Because no more than 25% of the cells have expected counts below 5, we can use the Chi-square test of association to test the following hypothesis.
• \(H_0:\) There is no association in the population between FacilityCondition and FacilityAge.
• \(H_A:\) There is an association in the population between FacilityCondition and FacilityAge.

chi2_con_age <- chisq.test(table(df$FacilityCondition, df$FacilityAge))
chi2_con_age

## 
##  Pearson's Chi-squared test
## 
## data:  table(df$FacilityCondition, df$FacilityAge)
## X-squared = 1660.1, df = 16, p-value < 2.2e-16

Because \(p\)-value < 2.2e-16 < \(\alpha\) = 0.05 and , we reject \(H_0\). As a result, the test is statistically significant to support \(H_A:\) There is an association in the population between FacilityCondition and FacilityAge.

Discussion

• Findings:
  • There is no association in the population between FacilityCondition and StreetType.
  • There is an association in the population between FacilityCondition and FacilityAge.
• Advantages:
  • The source is credible because it is coming from the government.
  • By using hypothesis testing, the result is statistically proved.
  • The dataset is large enough.
• Limitations:
  • Have not tested any exact association between FacilityCondition and FacilityAge.
  • Only about half facilities had the condition record.
  • This data set is limited to Victoria and does not take into account the situation in other states.
  • Only tested two possible factors (StreetType and FacilityAge).

Discussion Cont.

• Future investigations:
  • Test other possible factors. \(e.g.\) Difference of facilities in metropolitan areas and facilities in rural areas.
  • Investigate other states.
  • Try to find the exact relation between FacilityCondition and FacilityAge.