J. Bobadilla et al/ Knowledge-Based Systems 24(2011)1310-1316 1315 一 BestA· WorstGA -COR·cos-Ms b BestGA--WorstGA-COR--COS-MSD 125 e9o ♀虽§月导导爵忌昌員員 8員8月爵导导昌昌昌員員昌 c BestGA - WorstGA -COR--COS-MSD d 一 BestA· WorstGA=coR aawm Number of Recommendations Number of Recommendations Fig. 2. Traditional metrics and proposed genetic similarity method comparative results using FilmAffinity: (a) accuracy (Mean Absolute Error). (b)coverage. (c)precision, (d) -BestGA --WorstGA -COR--COS- MSD 一 BestA· WorstGA - COR·cos·MsD 2500 20.00 500 1000 員月呙导导呙昌昌員員曷 月員8呙导导昌国昌員員昌 BestGA .-WorstGA=coR·cos·MsD BestGA --- COR--COS-MSD 5 30 Number of recommendations Number of recommendations Fig 3. Traditional metrics and proposed genetic similarity method comparative results using Netflix: (a)accuracy(Mean Absolute Error). (b)coverage, (c)precision, (d)recall. 3 5.2. Performan Performance results obtained using: correlation, cosine, MSD and the proponed method Table 3 shows the results obtained in the experiments de- MSD GA-method scribed in Section 4.2. As may the time needed to make Time(s) predictions when using our GA is 42. 44% lower than the time needed when using Pearson correlation.5.2. Performance results Table 3 shows the results obtained in the experiments de￾scribed in Section 4.2. As may be seen, the time needed to make predictions when using our GA method is 42.44% lower than the time needed when using Pearson correlation. Fig. 2. Traditional metrics and proposed genetic similarity method comparative results using FilmAffinity: (a) accuracy (Mean Absolute Error), (b) coverage, (c) precision, (d) recall. Fig. 3. Traditional metrics and proposed genetic similarity method comparative results using Netflix: (a) accuracy (Mean Absolute Error), (b) coverage, (c) precision, (d) recall. Table 3 Performance results obtained using: correlation, cosine, MSD and the proponed GA￾method. Metric Correlation Cosine MSD GA-method Time (s) 28, 56 29, 19 25, 56 16, 47 J. Bobadilla et al. / Knowledge-Based Systems 24 (2011) 1310–1316 1315