|
Toggle Series |
||
|---|---|---|
| Waiting for file upload... | ||
| Waiting for file upload... | ||
| Waiting for file upload... | ||
| Waiting for file upload... | ||
| Waiting for file upload... | ||
|
Create Series | |
|---|---|
| Color: | |
| Width: | |
|
Regression Analysis | |
|---|---|
| Iterations: | |
| Name | R^2 |
N/A |
|
N/A |
|
N/A |
|
|
Create new series |
||
|---|---|---|
| aoX | average of X | The top and bottom 5% of solves will be removed from the average. The number of solves that will be removed is rounded up, so you cannot have an average of 2 or less. |
| moX | mean of X | Computes the mean of the last X solves, with every solve considered. |
|
Regression Analysis Mode |
|
|---|---|
| Iterations | How many iterations to run the regression for. ONLY AFFECTS THE POWER LAW REGRESSION More iterations will result in a more accurate regression, but will take longer. |
| R^2 | A measure from 0 to 1 of how well the regression explains the data. For a very bad fit, this can be negative. This is only computed after running the regression to avoid lag. |
| Power Law |
Fits data to y = A * N^(-B) + C using gradient descent. Will often overestimate the time of early solves, and may take a few seconds to compute. Power law of practice |
| Log-Log |
Fits data to y = A * N^B using regression. This is often the best regression. Will appear as a straight line when graphed on log-log axes. |
| Exponential |
Fits data to y = A * e^(N*B) using regression. This regression is often too flat. Will appear as a straight line when graphed with a log y-axis |
|
Overlay distribution |
||
|---|---|---|
| Button Name | Distribution Name | Parameter Estimation |
| Normal | Normal / Gaussian distribution | Maximum Likelihood |
| Skew | Skew normal distribution | Method of Moments |
| Beta | Beta distribution | Method of Moments |
| Gamma | Gamma distribution | Maximum Likelihood |
| Logit | Logit-normal distribution | Maximum Likelihood |
| Log | Log-normal distribution | Maximum Likelihood |
| KS: Kolmogorov-Smirnov test for goodness of fit. A lower number means a better fit | ||
| A^2: Anderson-Darling test for goodness of fit. A lower number means a better fit | ||
| Options: | ||
|---|---|---|
| Width: column width measured in seconds. Defaults to (round(log2(standard deviation/6)))^2 |
Window: The width of the sliding window. Defaults to 1/5 of the solves in the dataset |
Step: number of solves to move forward each frame Defaults to 1/500 of the solves in the dataset |
| X max: maximum value displayed on the x-axis Defaults to mean+3*standard deviation+1 |
Time: additional delay to add between frames measured in milliseconds Defaults to 1. Increasing this number will slow down the animation by lowering the fps |
|
| THIS FUNCTION IS NOT RECOMMENDED FOR SESSIONS WITH LESS THAN ~1000 SOLVES | ||
| Options: | ||
|---|---|---|
| Width: column width measured in seconds. Defaults to (round(log2(standard deviation/6)))^2 |
Step: number of solves to move forward each frame Defaults to 1/1000 of the solves in the dataset |
X max: max value displayed on the x-axis Defaults to mean+6*standard deviation+1 |
Series Settings (Master)
Click to apply preset
Adjust width for all series: