Impact of Other Factors on Deterioration
This section aims to examine the relationship between various variables (for which data is available for the study) and track deterioration. The available variables can be categorised into two main groups (based on their types) including continuous and categorical variables. This section adopts different approaches to examine the relationship each variable has with track deterioration. For the purpose of the analysis in this section, “Gauge Value Change per Month” is calculated for inspection points that were not repaired and causal relationships are investigated in relation to “Gauge Value Change per Month”.
Continuous Variables
If a variable can take on any value between two specified values, it is called a continuous variable. In our dataset, the following variables are continuous.
- • “MGT” as the million gross tonne value for the track where the inspection point is located.
- • “Trips” which is the number of trips for the track where the inspection point is located.
- • “Curve Radius” as the radius of curve for the track on which the inspection point is located.
- • “Years since Instalment” which is the number of years since tracks were originally installed.
To examine the relationship between available continuous variables and “Gauge Value Change per Month”, correlation analysis is adopted. Table 2 shows the correlation values and their associated p-values. Higher correlation values indicate higher association between the variable and “Gauge Value Change per Month”. P-values also show if the associated correlation value is statistically significant or not. If a p-value is greater than 0.05 (95% accuracy level), it means that the correlation value is not statistically significant. As seen in Table 2, MGT is positively correlated with “Gauge Value Change per Month”, and the p-value is 0 that means it is statistically correlated. “Trips” has a p-value greater than 0.05 so it is not significantly correlated with “Gauge Value Change per Month”. Both “Curve Radius” and “Years since Instalment” variables are significantly and negatively correlated
Table 2. Correlation analysis between “Gauge Value Change per Month” and continuous variables
Variable |
Correlation Value |
P-Value |
MGT |
0.09 |
0.00 |
Trips |
0.02 |
0.37 |
Curve Radius |
-0.15 |
0.00 |
Years since Instalment |
-0.14 |
0.00 |
with “Gauge Value Change per Month”. Their negative values indicate that as they increase, “Gauge Value Change per Month” decreases.
To obtain further insights into the impact of various variables on deterioration rate (i.e. “Gauge Value Change per Month”) Table 3 shows the average of various factors in different quartiles of “Gauge Value Change per Month”.
Also, Table 4 shows the values of different variables in different quartiles of “Curve Radius”. As seen, the greatest average “Gauge Value Change per Month”
Table 3. Average of various factors in different quartiles of “Gauge Value Change per Month”
Quartiles based on “Gauge Value Change per Month” |
“Gauge Value Change per Month” |
Trips |
MGT |
Years since Instalment |
Curve Radius |
1 (0.00 - 0.07) |
0.04 |
52,384 |
1,486,005 |
24 |
142 |
2 (0.08 - 0.15) |
0.11 |
49,226 |
1,410,567 |
27 |
163 |
3 (0.16 - 0.30) |
0.22 |
47,030 |
1,371,649 |
28 |
183 |
4 (0.30 - 7.27) |
0.63 |
50,201 |
1,544,150 |
20 |
132 |
Total |
0.25 |
49,678 |
1,452,837 |
25 |
155 |
Table 4. Average of various factors in different quartiles of "Curve Radius”
Quartiles based on “Curve Radius” |
Curve Radius |
“Gauge Value Change per Month” |
Trips |
MGT |
Years since Instalment |
1 |
18 |
0.19 |
73,343 |
2,096,975 |
9 |
2 |
61 |
0.41 |
50,457 |
1,625,048 |
12 |
3 |
229 |
0.17 |
33,430 |
804,620 |
53 |
4 |
268 |
0.20 |
44,220 |
1,289,104 |
30 |
Total |
155 |
0.25 |
49,678 |
1,452,837 |
25 |
occurs in the second quartile where the average radius is around 61m. This group consists of tracks that are relatively new (aged 12 years that is almost half of the average age of 25 years) experiencing almost average number of “Trips” and “MGT” compared to the overall “Trips” and “MGT” values across the entire sample. This contradicts our expectations, and therefore we hypothesise that there are other factors causing degradation in this group of tracks.