def testHW2_subset(): # Success
test, train = utils.load_and_normalize_housing_set()
df_full = pd.DataFrame(train)
df_test = utils.train_subset(df_full, ['CRIM', 'TAX', 'B', 'MEDV'], n=10)
df_train = utils.train_subset(df_full, ['CRIM', 'TAX', 'B', 'MEDV'], n=10)
dfX_test = pd.DataFrame([df_test['CRIM'], df_test['TAX'], df_test['MEDV']]).transpose()
dfX_train = pd.DataFrame([df_train['CRIM'], df_train['TAX'], df_train['MEDV']]).transpose()
print hw2.linear_gd(dfX_train, dfX_test, 'MEDV')
def testHW2_allcols(): # Fail
test, train = utils.load_and_normalize_housing_set()
df_full = pd.DataFrame(train)
cols = [col for col in df_full.columns if col != 'MEDV']
df_test = utils.train_subset(df_full, cols, n=10)
df_train = utils.train_subset(df_full, cols, n=10)
#dfX_test = pd.DataFrame([df_test['CRIM'], df_test['TAX'], df_test['MEDV']]).transpose()
#dfX_train = pd.DataFrame([df_train['CRIM'], df_train['TAX'], df_train['MEDV']]).transpose()
print hw2.linear_gd(df_train, df_test, 'MEDV')
def testPdToDict():
df = hw3.load_and_normalize_spambase()
cols = df.columns[0:3]
sub = utils.train_subset(df, cols, 5)
print sub
print hw3.pandas_to_data(sub)
def testTransposeArray():
dfup = hw3.load_and_normalize_spambase()
cols = dfup.columns[0:3]
sub = utils.train_subset(dfup, cols, 5)
up = hw3.pandas_to_data(sub)
print up
trans = hw3.transpose_array(up)
print trans
def testScale():
test, train = utils.load_and_normalize_housing_set()
df_full = pd.DataFrame(train)
df = utils.train_subset(df_full, ['CRIM', 'TAX', 'B', 'MEDV'], n=10)
w = []
for i in range(0,len(df['TAX'])):
w.append(random.random())
scaled = utils.scale(w, min(df['TAX']), max(df['TAX']))
plot.fit_v_point([w, df['MEDV'], scaled])
def testGradient_by_columns(df, cols): # fail
df = utils.train_subset(df, cols, n=len(df))
#dfX = pd.DataFrame([df['CRIM'], df['TAX']]).transpose()
print len(df)
print df
#raw_input()
fit = gd.gradient(df, df['MEDV'].head(len(df)), .00001, max_iterations=5000)
print 'read v fit'
print len(df)
print df['MEDV'].head(10)
print fit
print np.dot(df, fit)
def testGradient(): # Great success with subset
test, train = utils.load_and_normalize_housing_set()
df_full = pd.DataFrame(train)
subset_size = 100
df = utils.train_subset(df_full, ['CRIM', 'TAX', 'B', 'MEDV'], n=subset_size)
dfX = pd.DataFrame([df['CRIM'], df['TAX']]).transpose()
print len(dfX)
print dfX
#raw_input()
fit = gd.gradient(dfX, df['MEDV'].head(subset_size), .5, max_iterations=300)
print 'read v fit'
print len(dfX)
print df['MEDV'].head(10)
print fit
data = gd.add_col(gd.pandas_to_data(dfX), 1)
print np.dot(data, fit)
