Appending to a numpy array

Question

I have the following code:

result = np.zeros((samples,), dtype=[('time', '<f8'), ('data', '<f8', (datalen,))])

I would like to create variable tempresult that accumulates the data result, and once I have accumulated 25000 samples, I would like to perform some operation on it.

So I would like to do something like:

result = np.zeros((samples,), dtype=[('time', '<f8'), ('data', '<f8', (datalen,))])

tempresult.append(result)

if ( len(tempresult[0]  > 25000 )):
   # do something

I tried the answer code but I get exception TypeError: invalid type promotion

result = np.zeros((samples,), dtype=[('time', '<f8'), ('data', '<f8', (datalen,))])

        if self.storeTimeStamp:
            self.storeTimeStamp = False
            self.timestamp = message.timestamp64
            self.oldsamples = 0

        for sample in range(0, samples):
            sstep = self.timestamp + (self.oldsamples + sample) * step
            result[sample] = (sstep, data[sample])

        self.oldsamples = self.oldsamples + samples


        # append
        np.append(self.tempresult, result)

        if len(self.tempresult) < 25000:
            return
        return [self.tempresult]

If you are appending to self.tempresult, then len(self.tempresult) is what is increasing each time you do so. len(self.tempresult[0]) won't change (assuming that it's valid at all), and len(self.tempresult[0] > 25000 ) is just meaningless, you're asking for the length of a comparison result. — jasonharper
– jasonharper, Commented Aug 3, 2018 at 13:39
@jasonharper how would I initalize self.tempresult ? to None ? — andre
– andre, Commented Aug 3, 2018 at 13:41
Do not use np.append without first reading and understanding its python code. — hpaulj
– hpaulj, Commented Aug 3, 2018 at 14:54
Why aren't you using list append? stackoverflow.com/q/51665905/901925 — hpaulj
– hpaulj, Commented Aug 3, 2018 at 15:30

hpaulj · Accepted Answer · 2018-08-03 16:20:07Z

1

1) read np.append docs.

np.append(self.tempresult, result)

is wrong. np.append returns a new array; it does not act in place like list append.

2) np.append is a clumsy interface to np.concatenate. If you don't understand concatenate, you'll get messed up by append.

3) because it makes a new array each time, repeated concatenate is slow. It's much faster to collect a list of arrays, and do one concatenate at the end

4) when using a compound dtype, all inputs to concatenate have to have the same dtype.

answered Aug 3, 2018 at 16:20

hpaulj

233k14 gold badges260 silver badges392 bronze badges

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

Collectives™ on Stack Overflow

Appending to a numpy array

1 Answer 1

Comments

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Linked

Related