Xtest = X[trainlen+1:]
Yhtest = Yh[trainlen+1:]
Ystest = Ys[trainlen+1:]
Yvtest = Yv[trainlen+1:]
idstest = ids[trainlen+1:]
rh = regression_one(Xtrain, Yhtrain)
rs = regression_one(Xtrain, Ystrain)
rv = regression_one(Xtrain, Yvtrain)
Yhpred = np.sum(Xtest * rh, axis=1)
Yspred = np.sum(Xtest * rs, axis=1)
Yvpred = np.sum(Xtest * rv, axis=1)
h_avg = average_vector_dist(Yhpred, h_dict, idstest)
s_avg = average_vector_dist(Yspred, s_dict, idstest)
v_avg = average_vector_dist(Yvpred, v_dict, idstest)
avg = (h_avg + s_avg + v_avg)/3
if (avg < best_avg):
best_avg = avg
best_h = h_avg
best_s = s_avg
best_v = v_avg
print "BEST"
print "h: " + str(best_h)
print "s: " + str(best_s)
print "v: " + str(best_v)
Ybtrain = Yb[1:trainlen]
Xtest = X[trainlen + 1:]
Yrtest = Yr[trainlen + 1:]
Ygtest = Yg[trainlen + 1:]
Ybtest = Yb[trainlen + 1:]
idstest = ids[trainlen + 1:]
rr = regression_one(Xtrain, Yrtrain)
rg = regression_one(Xtrain, Ygtrain)
rb = regression_one(Xtrain, Ybtrain)
Yrpred = np.sum(Xtest * rr, axis=1)
Ygpred = np.sum(Xtest * rg, axis=1)
Ybpred = np.sum(Xtest * rb, axis=1)
r_avg = average_vector_dist(Yrpred, r_dict, idstest)
g_avg = average_vector_dist(Ygpred, g_dict, idstest)
b_avg = average_vector_dist(Ybpred, b_dict, idstest)
avg = (r_avg + g_avg + b_avg) / 3
if (avg < best_avg):
best_avg = avg
best_r = r_avg
best_g = g_avg
best_b = b_avg
print "BEST"
print "r: " + str(best_r)
print "g: " + str(best_g)
print "b: " + str(best_b)
Yarotrain = Yaro[1:trainlen]
Xtest = X[trainlen+1:]
Yvatest = Yva[trainlen+1:]
Yarotest = Yaro[trainlen+1:]
idstest = ids[trainlen+1:]
#print Yarotest[0]
rva = regression_one(Xtrain, Yvatrain)
raro = regression_one(Xtrain, Yarotrain)
Yvapred = np.sum(Xtest * rva, axis=1)
Yaropred = np.sum(Xtest * raro, axis=1)
va_avg = average_vector_dist(Yvapred, va_dict, idstest)
aro_avg = average_vector_dist_polar(Yaropred, aro_dict, idstest)
avg = (va_avg + aro_avg)/2
if (avg < best_avg):
best_avg = avg
best_va = va_avg
best_aro = aro_avg
print "BEST"
print "r: " + str(best_va)
print "theta: " + str(best_aro)
Yvtrain = Yv[1:trainlen]
Xtest = X[trainlen + 1:]
Yhtest = Yh[trainlen + 1:]
Ystest = Ys[trainlen + 1:]
Yvtest = Yv[trainlen + 1:]
idstest = ids[trainlen + 1:]
rh = regression_one(Xtrain, Yhtrain)
rs = regression_one(Xtrain, Ystrain)
rv = regression_one(Xtrain, Yvtrain)
Yhpred = np.sum(Xtest * rh, axis=1)
Yspred = np.sum(Xtest * rs, axis=1)
Yvpred = np.sum(Xtest * rv, axis=1)
h_avg = average_vector_dist(Yhpred, h_dict, idstest)
s_avg = average_vector_dist(Yspred, s_dict, idstest)
v_avg = average_vector_dist(Yvpred, v_dict, idstest)
avg = (h_avg + s_avg + v_avg) / 3
if (avg < best_avg):
best_avg = avg
best_h = h_avg
best_s = s_avg
best_v = v_avg
print "BEST"
print "h: " + str(best_h)
print "s: " + str(best_s)
print "v: " + str(best_v)
Xtrain = X[1:trainlen]
Yvatrain = Yva[1:trainlen]
Yarotrain = Yaro[1:trainlen]
Xtest = X[trainlen+1:]
Yvatest = Yva[trainlen+1:]
Yarotest = Yaro[trainlen+1:]
idstest = ids[trainlen+1:]
rva = regression_one(Xtrain, Yvatrain)
raro = regression_one(Xtrain, Yarotrain)
Yvapred = np.sum(Xtest * rva, axis=1)
Yaropred = np.sum(Xtest * raro, axis=1)
va_avg = average_vector_dist(Yvapred, va_dict, idstest)
aro_avg = average_vector_dist(Yaropred, aro_dict, idstest)
avg = (va_avg + aro_avg)/2
if (avg < best_avg):
best_avg = avg
best_va = va_avg
best_aro = aro_avg
print "BEST"
print "v: " + str(best_va)
print "a: " + str(best_aro)
Xtest = X[trainlen+1:]
Yrtest = Yr[trainlen+1:]
Ygtest = Yg[trainlen+1:]
Ybtest = Yb[trainlen+1:]
idstest = ids[trainlen+1:]
rr = regression_one(Xtrain, Yrtrain)
rg = regression_one(Xtrain, Ygtrain)
rb = regression_one(Xtrain, Ybtrain)
Yrpred = np.sum(Xtest * rr, axis=1)
Ygpred = np.sum(Xtest * rg, axis=1)
Ybpred = np.sum(Xtest * rb, axis=1)
r_avg = average_vector_dist(Yrpred, r_dict, idstest)
g_avg = average_vector_dist(Ygpred, g_dict, idstest)
b_avg = average_vector_dist(Ybpred, b_dict, idstest)
avg = (r_avg + g_avg + b_avg)/3
if (avg < best_avg):
best_avg = avg
best_r = r_avg
best_g = g_avg
best_b = b_avg
print "BEST"
print "r: " + str(best_r)
print "g: " + str(best_g)
trainlen = int(len(Yaro) * 0.7)
Xtrain = X[1:trainlen]
Yvatrain = Yva[1:trainlen]
Yarotrain = Yaro[1:trainlen]
Xtest = X[trainlen + 1:]
Yvatest = Yva[trainlen + 1:]
Yarotest = Yaro[trainlen + 1:]
idstest = ids[trainlen + 1:]
#print Yarotest[0]
rva = regression_one(Xtrain, Yvatrain)
raro = regression_one(Xtrain, Yarotrain)
Yvapred = np.sum(Xtest * rva, axis=1)
Yaropred = np.sum(Xtest * raro, axis=1)
va_avg = average_vector_dist(Yvapred, va_dict, idstest)
aro_avg = average_vector_dist_polar(Yaropred, aro_dict, idstest)
avg = (va_avg + aro_avg) / 2
if (avg < best_avg):
best_avg = avg
best_va = va_avg
best_aro = aro_avg
print "BEST"
print "r: " + str(best_va)
print "theta: " + str(best_aro)