Thermostat Reality Check

A building’s thermostat sometimes spits out impossible readings below L or above U. Before your system compares readings with k-NN, it first pulls them back to reality so neighbors are judged fairly.

You are asked to read an initial set of readings with targets, then answer queries using k-NN Regression, where all features are confined to [L, U] before comparison. Predictions are the mean of the k nearest targets under the standard Euclidean distance. If several candidates are at the same distance around the cutoff, pick the item that appears earlier in the input. You may assume n ≥ k.

The first line of the input contains five values n d k L U. Each of the next n lines contains d floating-point numbers followed by a floating-point target.

The next line contains a single integer q representing how many new readings to predict. Each of the next q lines contains d floating-point numbers to compare against the initial set after applying the same clipping rule.

The program should print q lines. Each line should contain a single floating-point number - the predicted target.