Table 6.
Feature Name |
ANOVA
R2 (%) |
Linear
Regression R2 (%) |
Quadratic Regression
R2 (%) |
Variance explained
(%) |
Regression Equation |
---|---|---|---|---|---|
somaCount | 65.9 | 18.6 | 62.7 | 95.1 | - 35.3x2 + 80.4x + 219.3 |
somaArea | 62.3 | 7.3 | 45.3 | 72.7 | 13400x2 + 45000x + 5200 |
neurite Length | 94.4 | 91.5 | 91.7 | 96.9 | - 26100x + 81800 |
neuriteArea | 95.3 | 91.2 | 91.6 | 95.6 | - 63500x + 213000 |
attachment Point# | 69.5 | 26.6 | 66.7 | 95.9 | - 212x2 + 449x + 1230 |
endingPoint# | 82.5 | 82.3 | 82.3 | 99.7 | - 976x + 4420 |
branchPoint# | 91.0 | 88.1 | 89.7 | 96.8 | 203x2 + 2080x + 5030 |
Avg_soma Area | 76.8 | 70.4 | 70.4 | 91.6 | 69.4x + 238 |
Avg_neurite Length | 88.0 | 82.8 | 86.1 | 97.8 | 22.4x2 - 177x + 381 |
Avg_neurite Area | 91.3 | 87.3 | 88.3 | 95.6 | 29.7x2 - 352x + 946 |
Avg_attachment Point# | 58.4 | 42.3 | 56.9 | 97.4 | - 0.167x2 + 0.207x + 5.60 |
Avg_ending Point# | 68.6 | 54.2 | 68.0 | 99.1 | 1.77x2 - 8.62x + 20.5 |
Avg_branch Point# | 85.1 | 75.7 | 84.0 | 98.7 | 2.10x2 - 12.4x + 23.1 |
The R2 ANOVA column shows the total amount of between group variation for each feature. The next two columns show how much of the total variation is explained by a linear model or a quadratic model. If the quadratic model explains more than 5% of the remaining between group variation over the linear model that model is chosen otherwise the quadratic model doesn't improve the fit significantly enough to justify the additional complexity. Essentially, if using the quadratic model increases the R2 fit by more than 0.5% over the linear model we choose that one [35] The variance explained column shows the total percentage of the between group variation explained by the chosen model. The final column shows the equation for the best fit model for each feature.