Create educational notebook based on Sargolini 2006

[weiglszonja]

Developer Days Project

[bendichter]

@weiglszonja what still needs to be done with this PR? Is it ready for review?

[weiglszonja]

@bendichter Adding the cell layer information is missing currently, I checked the file that contains this, should not be hard to match it with the existing data. I can do it this week.

[weiglszonja]

@bendichter ready for review

[bendichter]

What is the purpose of

def pearson_cor_2arr(arr1, arr2):
    """ Calculating Pearson correlation for two arrays with the same shape and no NaN."""
    if not arr1.shape == arr2.shape:
        raise Exception("Both arrays should have the same shape")
    n = arr1.shape[0] * arr1.shape[1]
 
    numerator = n * np.sum(arr1 * arr2) - np.sum(arr1) * np.sum(arr2)
    denominator = (n * np.sum(arr1**2) - np.sum(arr1)**2)**0.5 * (n * np.sum(arr2**2) - np.sum(arr2)**2)**0.5
    return numerator / denominator

I think this should be the same as scipy.stats.pearsonr(arr1[:], arr2[:])[0]

[weiglszonja]

What is the purpose of

def pearson_cor_2arr(arr1, arr2):
    """ Calculating Pearson correlation for two arrays with the same shape and no NaN."""
    if not arr1.shape == arr2.shape:
        raise Exception("Both arrays should have the same shape")
    n = arr1.shape[0] * arr1.shape[1]
 
    numerator = n * np.sum(arr1 * arr2) - np.sum(arr1) * np.sum(arr2)
    denominator = (n * np.sum(arr1**2) - np.sum(arr1)**2)**0.5 * (n * np.sum(arr2**2) - np.sum(arr2)**2)**0.5
    return numerator / denominator

I think this should be the same as scipy.stats.pearsonr(arr1[:], arr2[:])[0]

You're right, the arrays just have to be flattened, I don't remember why we didn't use pearsonr (let me know @djarecka if you do). Changed it in the notebook

[yarikoptic]

looks good to me. @bendichter do you give your blessing ? then please click big green button!

[djarecka]

What is the purpose of

def pearson_cor_2arr(arr1, arr2):
    """ Calculating Pearson correlation for two arrays with the same shape and no NaN."""
    if not arr1.shape == arr2.shape:
        raise Exception("Both arrays should have the same shape")
    n = arr1.shape[0] * arr1.shape[1]
 
    numerator = n * np.sum(arr1 * arr2) - np.sum(arr1) * np.sum(arr2)
    denominator = (n * np.sum(arr1**2) - np.sum(arr1)**2)**0.5 * (n * np.sum(arr2**2) - np.sum(arr2)**2)**0.5
    return numerator / denominator

I think this should be the same as scipy.stats.pearsonr(arr1[:], arr2[:])[0]

You're right, the arrays just have to be flattened, I don't remember why we didn't use pearsonr (let me know @djarecka if you do). Changed it in the notebook

@weiglszonja - sorry, missed the question. Don't remember, why I created the function on my own, perhaps had some issues with the scipy function or wanted to compare. we can check if this gives the same results now.

[weiglszonja]

@djarecka no worries, I checked already.

[CodyCBakerPhD]

@weiglszonja @bendichter Any updated on this? Is this ready?

[weiglszonja]

@CodyCBakerPhD This is good to go from my side, but I don't have permission to merge.