If I use F1 score as a metric, that classifier is going to get a low score. My classifier ignores the input and always returns the same prediction: “has flu.” The recall of this classifier is going to be 1 because I correctly classified all sick patients as sick, but the precision is near 0 because of a considerable number of false positives. Such a function is a perfect choice for the scoring metric of a classifier because useless classifiers get a meager score.įor example, if I created a fake “classifier” that tells a doctor whether a patient has the flu or not. Print(stats.hmean()) # added one extremely small value Print(stats.hmean()) # added one small value Print(stats.hmean()) #added one extremely large value Print(stats.hmean()) #added one large value Extremely low values have a significant influence on the result. On the other hand, not all outliers are ignored. Because of that, the result is not sensitive to extremely large values. The harmonic mean is defined as the reciprocal of the arithmetic mean of the reciprocals. The F1 score is based on the harmonic mean. If one of the parameters is small, the second one no longer matters.Īs I mentioned at the beginning, F1 score emphasizes the lowest value. It looks that in this case precision is ignored, and the F1 score remain equal to 0. Let’s begin by looking at extreme values.įor example precision = 1 and recall = 0. Plot(all_values, all_values, all_values, f_score_label = 'F1 score') Sc = ax2.scatter(precision, recall, c = f_score, cmap=plt.cm.jet, vmin = 0, vmax = 1) It's been relatively closely matched at Haas, where Kevin Magnussen has given Romain Grosjean much more competition than Esteban Gutierrez in 2016.Def plot(precision, recall, f_score, f_score_label):Īx = plt.subplot(gs, projection='3d')Īx.plot_trisurf(precision, recall, f_score, cmap=plt.cm.jet, linewidth=0.2, vmin = 0, vmax = 1)Īx.set_zlabel(f_score_label, rotation = 0) Head to head: Hulkenberg 15-0 Palmer, Hulkenberg 3-1 SainzĪverage gap: Hulkenberg -0.738 Palmer, Hulkenberg -0.198s Sainz Palmer, while hamstrung by reliability, also suffered the widest average gap to a team-mate on the grid, and Carlos Sainz has immediately got closer to Hulkenberg in his few attempts. The most comprehensive qualifying record, by far, went in Hulkenberg's favour at Renault, with Jolyon Palmer failing to match the German even once. Head to head: Sainz 8-6 Kvyat, Gasly 1-1 HartleyĪverage gap: Sainz -0.114s Kvyat, Gasly -0.343s Hartley Pierre Gasly then replaced the Russian, and the team have now settled on a Gasly- Brendon Hartley partnership for 2018. It's been a season of change at Toro Rosso.įirst, the highly-rated Carlos Sainz continued his superiority over Daniil Kvyat, though it was closer than many expected before Kvyat was unceremoniously dumped. Stroll beat Massa in Azerbaijan before a stunning podium finish, and also got the better of his team-mate when dazzling in the wet to start on the front row in Italy. However, it's fair to say Stroll's two qualifying victories have been memorable. Massa was beaten 17-4 by Bottas last year but has been utterly dominant in 2017 - with an average advantage of 0.6 seconds. Lance Stroll has struggled on Saturdays in his rookie F1 season next to an experienced campaigner in Felipe Massa at Williams. He beat Perez in four consecutive qualifying sessions prior to the penultimate race in Brazil - no mean feat. They have been separated by less than a tenth of a second six times.Īverage gap: Verstpapen -0.128s RicciardoĪfter many a tight battle with Nico Hulkenberg, Sergio Perez welcomed a new team-mate at Force India in 2017.Įsteban Ocon struggled at first but has certainly impressed in his first full season in F1 - and is getting better and better on a Saturday. Verstappen even enjoyed a six-weekend streak of intra-team victories ahead of the Japanese GP, though the two are still very evenly matched. Daniel Ricciardo may have scored more points than Max Verstappen at Red Bull, but it's the Dutchman who has reversed his 2016 deficit into a comfortable qualifying advantage this season.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |