summaryrefslogtreecommitdiff
path: root/run_tests.m
AgeCommit message (Collapse)AuthorFilesLines
2012-05-12opt_pred: Split opt_pred in opt_idx and opt_pred.Jan Klemkow1-1/+2
So you can use the idx of opt_idx for other predictions.
2012-05-12chi_pred: add an inactive chi rand. pred method.Jan Klemkow1-1/+2
2012-05-11run_tests: Disable all plotsBenjamin Franzke1-2/+2
2012-05-11run_tests: Pack struct definition into a anonymous functionBenjamin Franzke1-7/+8
2012-05-11run_tests: Use end-keyword for appending to a vectorBenjamin Franzke1-1/+1
2012-05-11run_tests: Use an vector of structs for the prediction methodsBenjamin Franzke1-11/+11
Allows use to use e.g. pred_methods.name.
2012-05-11run_tests: Optimize local_min_count storageBenjamin Franzke1-2/+2
2012-05-11run_tests: Add optimze errors into qerr/terr directlyBenjamin Franzke1-3/+1
2012-05-11Put local error calculation into calc_errorBenjamin Franzke1-6/+3
2012-05-11Store all non-emta prediction methods in a cell-arrayBenjamin Franzke1-30/+25
2012-05-11Move frequency removal regression into its own functionBenjamin Franzke1-7/+1
2012-05-11Remove log+quad algorithmBenjamin Franzke1-4/+1
Its really not needed since wehave optimize
2012-05-11sevenday_pred: Use same prototype as for other testsBenjamin Franzke1-1/+1
2012-05-11mean/rand_pred: Use same function prototype as regress_predBenjamin Franzke1-2/+2
The number of days is implicit in the size of the given price and quantity matrices.
2012-05-11run_tests: Store local_min_count in a variable as wellBenjamin Franzke1-2/+4
2012-05-11Print local min count for the prediction algorithmsBenjamin Franzke1-2/+7
2012-05-11Remove logarithmic + quad from optimized predictionBenjamin Franzke1-3/+3
..since optimized prediction does the same but more.
2012-05-11Add sevenday removal algorithm to optimize-predictionBenjamin Franzke1-12/+13
2012-05-11Add optimization method.Jan Klemkow1-2/+8
2012-05-10run_tests: rotate err matrics.Jan Klemkow1-10/+10
2012-05-10run_tests: calculate global min. error.Jan Klemkow1-1/+19
2012-05-10run_tests: Rather call the combined l&qBenjamin Franzke1-1/+1
..since its mainly logarithmic.
2012-05-10Combine logarithmic and quad predictionBenjamin Franzke1-0/+1
Product 335 scales better with quadratic prediction.
2012-05-10calc_error: return quad and total errors.Jan Klemkow1-6/+6
2012-05-10run_tests: Code clean up.Jan Klemkow1-6/+16
2012-05-10Add quadratic interterpolationBenjamin Franzke1-0/+2
Not usable at all, worse than mean value.
2012-05-10Add logarithmic interpolation predectionBenjamin Franzke1-0/+1
Slightly better than linear regression.
2012-05-10Add a method for random prediction.Jan Klemkow1-0/+1
This method show the distace from nonsence to the other methods. It uses the variance of each product for prediction.
2012-05-08Add frequency removal to run_tests (bad result)Benjamin Franzke1-0/+8
2012-05-07Display type in run_testsJan Klemkow1-4/+4
2012-05-07Calculate errors in seperate function and plot error graph.Jan Klemkow1-13/+6
2012-05-07Add sevenday prediction.Jan Klemkow1-4/+10
2012-05-07Add total error calculation.Jan Klemkow1-2/+9
2012-04-24Add prediction test using linear regressionBenjamin Franzke1-0/+1
2012-04-24run_tests.m: Initialize product dataJan Klemkow1-0/+1
2012-04-24Add mean prediction testJan Klemkow1-0/+1