Udacity CarND Term3 3. Prediction exercise implementing naive Bayes.
X_train number of elements 750
X_train element size 4
Y_train number of elements 750
X_test number of elements 250
X_test element size 4
Y_test number of elements 250
**You got 84.80000 correct**
- Feature 1: even occuring probability:
p_left=0.28533, p_keep=0.42133, p_right=0.29333
- Feature 2: offset from road center. Notice keep deviation is really small besides its small offset.
left_d_offset_ave=0.73940, keep_d_offset_ave=0.17360, right_d_offset_ave=0.71776
left_d_offset_dev=0.30399, keep_d_offset_dev=0.09271, right_d_offset_dev=0.39838
- Feature 3: d_dot be negative/left, almost zero/keep, positive/right.
left_d_dot_ave=-0.96709, keep_d_dot_ave=0.00581, right_d_dot_ave=0.95402
left_d_dot_dev=0.43994, keep_d_dot_dev=0.02827, right_d_dot_dev=0.41841
- Prediction can be done using Gaussian Naive Bayes method.
- Wikipedia has a very good example of sex classification here: https://en.wikipedia.org/wiki/Naive_Bayes_classifier#Gaussian_naive_Bayes
- The theoretcal explanation is also good: https://en.wikipedia.org/wiki/Naive_Bayes_classifier#Parameter_estimation_and_event_models
- Use Gaussian distribution to emulate the probability. Need to get u and sigma of each feature.
// u = average over all same labelled data
// sigma^2 = sum((x_i - u)^2)/N
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent