Saucier self-rating / PDA matrix

Data: https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/GHYMEV

There was missing data, which I filled using the average over the rest of the scores.

First dimension

Second dimension

Third dimension

Fourth dimension

Fifth dimension

Sixth dimension

Seventh dimension

Eighth dimension

Ninth dimension

Tenth dimension

Eleventh dimension

Twelfth dimension

Thirteenth dimension

Fourteenth dimension

Fifteenth dimension

Sixteenth dimension

Seventeenth dimension

Eighteenth dimension

Nineteenth dimension

Twentieth dimension

The sigma values

First 20 sigma values

First 10 sigma values

So they drop off quickly!

The first 3 columns of U

The first 3 columns of U for the first 20 terms

To reproduce: Download the file as .tab from https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/GHYMEV and load it as a dataframe: trysauce = pd.read_csv("525_PDA-1.tab",sep="\t") Drop the column that has the IDs of the participants: trysauce2 = trysauce.drop('ID',axis=1) Subtract the mean: trysauce3 = trysauce2 - trysauce2.mean().mean() Find missing values: trysauce3.isnull().any() Fill missing values with the mean *note: I should have used the new mean after the mean was removed instead of the original mean: trysauce4 = trysauce3.fillna(trysauce2.mean().mean()) Run SVD: t3df, t3u, t3d, t3v, t3sig, t3x, t3remakex = runSVD(trysauce4)

The mean of the df after the original mean was removed:

m = trysauce3.mean().mean() = -9.42272714880909e-15
trysauce5 = trysauce3.fillna(m)
t3df, t3u, t3d,  t3v, t3sig, t3x, t3remakex = runSVD(trysauce5)

Rerunning SVD using that value doesn't really seem to make any difference (yay), so I'm not going to redo everything:

jwzimmer-zz / tv-tropening

Saucier self-rating / PDA matrix #19