python simple if else structure wrong comparison

Question

This is my code

for i,val in enumerate(DS3Y_pred_trans):
    if val < 1.5:
        DS3Y_pred_trans[i] = 1
    else:
        DS3Y_pred_trans[i] = 2

There are values less than 1.5 in the list, but the out is all 2s.

What am I missing?

This is the whole code.

from numpy import genfromtxt
DS3X_train = np.genfromtxt('train.csv', dtype=float, delimiter=',')
print DS3X_train

DS3Y_train = np.genfromtxt('train_labels.csv', dtype=int, delimiter=',' )
print DS3Y_train

DS3X_test = np.genfromtxt('test.csv', dtype=float, delimiter=',')
print DS3X_test

DS3Y_test = np.genfromtxt('test_labels.csv', dtype=int, delimiter=',' )
print DS3Y_test

DS3X_train_trans = zip(*DS3X_train)

cov_train = np.cov(DS3X_train_trans)
U, s, V = np.linalg.svd(cov_train, full_matrices=True)
u = U[:,:-1]
u_trans = zip(*u)
DS3X_train_reduced = np.dot(u_trans,DS3X_train_trans) 

b = np.ones((3,2000))
b[1:,:] = DS3X_train_reduced
print "\n"
DS3X_train_reduced = b


DS3X_train_reduced_trans = zip(*DS3X_train_reduced)

temp = np.dot(DS3X_train_reduced,DS3X_train_reduced_trans)

try:
    inv_temp = np.linalg.inv(temp)
except np.linalg.LinAlgError:

    pass
else:

    psue_inv = np.dot(inv_temp,DS3X_train_reduced)
    print psue_inv.shape

weight = np.dot(psue_inv,DS3Y_train)
weight_trans = zip(weight)
print weight_trans



DS3X_test_trans = zip(*DS3X_test)
DS3X_test_reduced = np.dot(u_trans,DS3X_test_trans)

b = np.ones((3,400))
b[1:,:] = DS3X_test_reduced
print "\n"
print b
DS3X_test_reduced = b

print DS3X_test_reduced.shape
DS3X_test_reduced_trans = zip(*DS3X_test_reduced)
DS3Y_pred = np.dot(DS3X_test_reduced_trans,weight_trans)
print DS3Y_pred
print DS3Y_pred.shape 

DS3Y_pred_trans = zip(DS3Y_pred)

print repr(DS3Y_pred_trans[0])

for i,val in enumerate(DS3Y_pred_trans):
    if val < 1.5:
        DS3Y_pred_trans[i] = 1
    else:
        DS3Y_pred_trans[i] = 2


print DS3Y_pred
now regression using indicator variable and graph plottings

Could not replicate, using e.g. DS3Y_pred_trans = [0, 0.5, 1, 1.5, 2, 2.5, 3] this works fine for me. What do you have in DS3Y_pred_trans ? — jonrsharpe
– jonrsharpe, Commented Jan 6, 2015 at 12:24
Almost certainly your values are strings, not numbers. In Python 2 numbers sort before strings, so val < 1.5 is then always going to be false. — Martijn Pieters
– Martijn Pieters, Commented Jan 6, 2015 at 12:25
How can i check if they are strings or numbers? Sorry i am new to python. — user2871856
– user2871856, Commented Jan 6, 2015 at 12:27
@user2871856: give us the output of print DS3Y_pred_trans; strings will be represented as Python string literals, e.g. with quotes. — Martijn Pieters
– Martijn Pieters, Commented Jan 6, 2015 at 12:28
@user2871856: then you have numpy arrays in your list; each element is another list-like object. How did you build this structure? — Martijn Pieters
– Martijn Pieters, Commented Jan 6, 2015 at 12:38

Martijn Pieters · Accepted Answer · 2015-01-06 12:39:13Z

5

Your values are not numbers. In Python 2 numbers sort before other objects, so when comparing val with 1.5, the comparison is always false.

You probably have strings:

>>> '1.0' < 1.5
False
>>> 1.0 < 1.5
True

If so, convert your values to floats first:

for i, val in enumerate(DS3Y_pred_trans):
    if float(val) < 1.5:
        DS3Y_pred_trans[i] = 1
    else:
        DS3Y_pred_trans[i] = 2

It could be that you are storing other objects in the list still; you'll need to take a close look at what is actually in the list and adjust your code accordingly, or fix how the list is created in the first place.

Since you are replacing all values anyway, you could use a list comprehension:

DS3Y_pred_trans = [1 if float(val) < 1.5 else 2 for val in DS3Y_pred_trans]

edited Jan 6, 2015 at 12:39

answered Jan 6, 2015 at 12:26

Martijn Pieters

1.1m326 gold badges4.2k silver badges3.4k bronze badges

Sign up to request clarification or add additional context in comments.

12 Comments

Padraic Cunningham Over a year ago

You don't need to provide a bad/wrong answer to get downvoted on SO.

Martijn Pieters Over a year ago

@PadraicCunningham: I am quite aware of that; but most people voting usually can motivate why, and if they are willing I'd love to hear that reasoning. Clearly they felt my answer was not helpful in some way.

Matthias Over a year ago

According to the newest comment from the OP it seems that we have to compare with val[0] (or val[0][0]- I couldn't quite see that).

user2871856 Over a year ago

its giving Type error .. float() argument must be string or a number.

Martijn Pieters Over a year ago

@user2871856: also, in future, you should really have included that sample from the start for us to reproduce your issues cleanly. I had to guess based on limited information, and now we end up with this huge long back and forth about what exactly you do have.

|

Collectives™ on Stack Overflow

python simple if else structure wrong comparison

1 Answer 1

12 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

12 Comments

Your Answer

Sign up or log in

Post as a guest

Related