Mean Rowwise Root Mean Squared Error

$\text{MMRMSE}=\frac{1}{R}{\sum }_{i=1}^R{\left(\frac{1}{n}{\sum }_{j=1}^n(y_{ij}-{\hat{y}}_{ij}{)}^2\right)}^{1/2}$

• 𝑅 is the number of scored rows,
• $y_{ij}$ and ${\hat{y}}_{ij}$ are the actual and predicted values for row i and column j.
• n is the number of columns

link: range: $[0,\infty )$

summary

Use when you have multiple targets to predict

• (which is why the submission is a matrix!)

pros

•  gives higher weight to larger errors (like RMSE)

Cons