Update GIN #87

fengxie009 · 2022-12-07T02:28:25Z

No description provided.

tofuwen

did part of review as code style examples~

tofuwen · 2022-12-07T02:50:37Z

causallearn/search/HiddenCausal/GIN/FisherTest.py

+#-------------------------------------------------------------------------------
+# Name:        模块1
+# Purpose:
+#
+# Author:      YY
+#
+# Created:     03/03/2021
+# Copyright:   (c) YY 2021
+# Licence:     <your licence>
+#-------------------------------------------------------------------------------


We should delete this, or fill it.

In the open-source code, we should maintain high code quality --- we shouldn't allow any unnecessary comments in the code.

Sorry there are lots of rules in industry code that are different from research code :(

tofuwen · 2022-12-07T02:52:08Z

causallearn/search/HiddenCausal/GIN/FisherTest.py

+import math
+from scipy.stats import chi2
+
+def FisherTest(pvals,alph=0.01):


nit: comma between different variables.

You can follow Google Python codestyle when writing code: https://google.github.io/styleguide/pyguide.html

You can find automated tool like pylint in the link.

Suggested change

def FisherTest(pvals,alph=0.01):

def FisherTest(pvals, alph=0.01):

tofuwen · 2022-12-07T02:52:24Z

causallearn/search/HiddenCausal/GIN/FisherTest.py

+from scipy.stats import chi2
+
+def FisherTest(pvals,alph=0.01):
+    Fisher_Stat=0


space between operand

Suggested change

Fisher_Stat=0

Fisher_Stat = 0

tofuwen · 2022-12-07T02:52:31Z

causallearn/search/HiddenCausal/GIN/FisherTest.py

+def FisherTest(pvals,alph=0.01):
+    Fisher_Stat=0
+    L = len(pvals)
+    for i in range(0,L):


Suggested change

for i in range(0,L):

for i in range(0, L):

address this?

tofuwen · 2022-12-07T02:52:39Z

causallearn/search/HiddenCausal/GIN/FisherTest.py

+    Fisher_Stat=0
+    L = len(pvals)
+    for i in range(0,L):
+        if pvals[i] ==0:


Suggested change

if pvals[i] ==0:

if pvals[i] == 0:

tofuwen · 2022-12-07T02:53:36Z

causallearn/search/HiddenCausal/GIN/FisherTest.py

+    if Fisher_pval >alph:
+        return True,Fisher_pval
+    else:
+        return False,Fisher_pval


Suggested change

if Fisher_pval >alph:

return True,Fisher_pval

else:

return False,Fisher_pval

return Fisher_pval > alph, Fisher_pval

tofuwen · 2022-12-07T02:53:57Z

causallearn/search/HiddenCausal/GIN/FisherTest.py

+
+
+def main():
+    pvals = [0.01,0.9]


Suggested change

pvals = [0.01,0.9]

pvals = [0.01, 0.9]

tofuwen · 2022-12-07T02:54:03Z

causallearn/search/HiddenCausal/GIN/FisherTest.py

+
+def main():
+    pvals = [0.01,0.9]
+    FisherTest(pvals,0.1)


Suggested change

FisherTest(pvals,0.1)

FisherTest(pvals, 0.1)

tofuwen · 2022-12-07T02:55:09Z

causallearn/search/HiddenCausal/GIN/GIN2.py

+        if pval > alpha:
+            flag = True
+        else:
+            flag = False
+
+        if not flag:#not false == ture  ---> if false
+            #print(X,Z,flag)
+            return False


Suggested change

if pval > alpha:

flag = True

else:

flag = False

if not flag:#not false == ture ---> if false

#print(X,Z,flag)

return False

if pval <= alpha:

return False

tofuwen · 2022-12-07T02:55:55Z

causallearn/search/HiddenCausal/GIN/HSIC2.py

+    if pvalue >alph:
+        return True
+    else:
+        return False


Suggested change

if pvalue >alph:

return True

else:

return False

return pvalue > alph

tofuwen · 2022-12-07T02:58:01Z

BTW, for every PR, we should have an informative description.

Could you add your description? Here is a good example PR with great descriptions: #85

fengxie009 · 2022-12-17T01:55:35Z

Updates

Modified the whole GIN package to make it more readable

GIN.py: GIN(data, indep_test_method='kci', alpha=0.05, MAX_Factor = 2)
The MAX_Factor parameter is added to make the GIN algorithm more efficient

Notes
Please kindly note that the GIN method highly relies on the independent test, e.g., KCI or HSIC. Please try to modify the default parameters in the independent test method, such as kernel width, when the GIN method can not output any information.

tofuwen

Thanks so much for your great work! This version is much better than previous version!

Added more comments.

And there are two main changes we need for this PR:

could you write tests for GIN method to make sure GIN behaves as expected?
could you edit your first message to include the description of this PR? We need to have great descriptions for every PR, so in the future, when we come back or if our users are checking the PR, we / they can know what is happening.

tofuwen · 2022-12-20T20:29:32Z

causallearn/search/HiddenCausal/GIN/FisherTest.py

+import math
+from scipy.stats import chi2
+
+def FisherTest(pvals, alpha = 0.01):


Suggested change

def FisherTest(pvals, alpha = 0.01):

def FisherTest(pvals, alpha=0.01):

tofuwen · 2022-12-20T20:29:46Z

causallearn/search/HiddenCausal/GIN/FisherTest.py

+def FisherTest(pvals,alph=0.01):
+    Fisher_Stat=0
+    L = len(pvals)
+    for i in range(0,L):


address this?

tofuwen · 2022-12-20T20:30:07Z

causallearn/search/HiddenCausal/GIN/FisherTest.py

+        else:
+            TP = pvals[i]
+
+        Fisher_Stat = Fisher_Stat - 2*math.log(TP)


Suggested change

Fisher_Stat = Fisher_Stat - 2*math.log(TP)

Fisher_Stat = Fisher_Stat - 2 * math.log(TP)

tofuwen · 2022-12-20T20:30:16Z

causallearn/search/HiddenCausal/GIN/FisherTest.py

+
+        Fisher_Stat = Fisher_Stat - 2*math.log(TP)
+
+    Fisher_pval = 1 - chi2.cdf(Fisher_Stat, 2*L)


Suggested change

Fisher_pval = 1 - chi2.cdf(Fisher_Stat, 2*L)

Fisher_pval = 1 - chi2.cdf(Fisher_Stat, 2 * L)

tofuwen · 2022-12-20T20:30:50Z

causallearn/search/HiddenCausal/GIN/FisherTest.py

+    if Fisher_pval >alpha:
+        return True, Fisher_pval
+    else:
+        return False, Fisher_pval


Suggested change

if Fisher_pval >alpha:

return True, Fisher_pval

else:

return False, Fisher_pval

return Fisher_pval > alpha, Fisher_pval

tofuwen · 2022-12-20T20:37:21Z

causallearn/search/HiddenCausal/GIN/Mutual_Information.py

+def _entropy(u):
+    """Calculate entropy using the maximum entropy approximations."""
+    k1 = 79.047
+    k2 = 7.4129
+    gamma = 0.37457
+    return (1 + np.log(2 * np.pi)) / 2 - k1 * (np.mean(np.log(np.cosh(u))) - gamma)**2 - k2 * (np.mean(u * np.exp((-u**2) / 2)))**2


seems equivalent to previous functions?

Re-use code whenever possible and never duplicate code

tofuwen · 2022-12-20T20:37:44Z

causallearn/search/HiddenCausal/GIN/Mutual_Information.py

+
+def MI_1(x1, y1):
+    """estimating mutual information by Non-parametric computation of entropy and mutual-information"""
+    x=x1.reshape(len(x1),1)


Suggested change

x=x1.reshape(len(x1),1)

x = x1.reshape(len(x1), 1)

tofuwen · 2022-12-20T20:37:50Z

causallearn/search/HiddenCausal/GIN/Mutual_Information.py

+def MI_1(x1, y1):
+    """estimating mutual information by Non-parametric computation of entropy and mutual-information"""
+    x=x1.reshape(len(x1),1)
+    y=y1.reshape(len(y1),1)


tofuwen · 2022-12-20T20:39:13Z

causallearn/search/HiddenCausal/GIN/Test_GIN_Condition.py

+        if pval > alpha:
+            flag = True
+        else:
+            flag = False
+
+        if not flag:
+            return False
+
+    return True


I remembered I commented on this before?

Suggested change

if pval > alpha:

flag = True

else:

flag = False

if not flag:

return False

return True

if pval <= alpha:

return False

return True

tofuwen · 2022-12-20T20:39:38Z

causallearn/search/HiddenCausal/GIN/Test_GIN_Condition.py

+        temp = np.array(data[:, i])
+        mi = ID.MI_1(result, temp)
+        MIS += mi
+    MIS = MIS/len(Z)


Suggested change

MIS = MIS/len(Z)

MIS = MIS / len(Z)

tofuwen · 2022-12-20T20:43:44Z

Basically add description there. People will tend to read the first block, but not the conversions (because it's too much)

So in industry (or open-sourced project), it's a good practice to have great descriptions in the first cell.

kunwuz · 2023-01-31T00:10:34Z

Hi all @fengxie009 @tofuwen, just would like to quickly check whether there are any updates on this PR. Please feel free to let others know if there are any questions :)

fengxie009 · 2023-02-03T05:05:59Z

Hi Yujia, There are currently no updates. I will let you know if there are any changes. Best wishes, Feng Yujia Zheng ***@***.***> 于2023年1月31日周二 08:10写道：

…

Hi all @fengxie009 <https://github.com/fengxie009> @tofuwen <https://github.com/tofuwen>, just would like to quickly check whether there are any updates on this PR. Please feel free to let others know if there are any questions :) — Reply to this email directly, view it on GitHub <#87 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AH6YBRX2AGGKIDDJIR5YJPLWVBKAJANCNFSM6AAAAAASWHV7YU> . You are receiving this because you were mentioned.Message ID: ***@***.***>

tofuwen · 2023-03-22T23:08:59Z

Hi @fengxie009,

Do we have some updates here? :)

We need to ensure that we have unit tests coverage for all the methods, as it's required here: #100

And we will build the CI once your unit tests are completed. :)

kunwuz · 2023-10-12T23:29:37Z

Hi @fengxie009 and @tofuwen, are there any updates on this PR? I believe some quick tests to make sure that the modification is correct is enough to merge this. It seems to be around too long.

fengxie009 added 2 commits December 7, 2022 10:24

Delete GIN.py

a282d62

Add files via upload

89b80f1

tofuwen reviewed Dec 7, 2022

View reviewed changes

fengxie009 added 12 commits December 17, 2022 09:13

Delete FisherTest.py

fcbafd3

Add files via upload

200d833

Delete GIN.py

0db446d

Delete GIN2.py

9ad04fa

Delete HSIC2.py

1b2d4c7

Delete Identifying_Causal_Clusters.py

5229b06

Delete Infer_Causal_Order.py

ebb808b

Delete Utils_V2.py

9299ea2

Delete __init__.py

36d5d4b

Delete independence.py

c92cb8f

Delete mutual.py

4f168e8

Add files via upload

e307426

tofuwen reviewed Dec 20, 2022

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Update GIN #87

Update GIN #87

fengxie009 commented Dec 7, 2022

tofuwen left a comment

tofuwen Dec 7, 2022

tofuwen Dec 7, 2022

tofuwen Dec 7, 2022

tofuwen Dec 7, 2022

tofuwen Dec 20, 2022

tofuwen Dec 7, 2022

tofuwen Dec 7, 2022

tofuwen Dec 7, 2022

tofuwen Dec 7, 2022

tofuwen Dec 7, 2022

tofuwen Dec 7, 2022

tofuwen commented Dec 7, 2022

fengxie009 commented Dec 17, 2022

tofuwen left a comment

tofuwen Dec 20, 2022

tofuwen Dec 20, 2022

tofuwen Dec 20, 2022

tofuwen Dec 20, 2022

tofuwen Dec 20, 2022

tofuwen Dec 20, 2022

tofuwen Dec 20, 2022

tofuwen Dec 20, 2022

tofuwen Dec 20, 2022

tofuwen Dec 20, 2022

tofuwen commented Dec 20, 2022

kunwuz commented Jan 31, 2023

fengxie009 commented Feb 3, 2023 via email

tofuwen commented Mar 22, 2023

kunwuz commented Oct 12, 2023

	def FisherTest(pvals,alph=0.01):
	def FisherTest(pvals, alph=0.01):

	def FisherTest(pvals, alpha = 0.01):
	def FisherTest(pvals, alpha=0.01):

	Fisher_Stat = Fisher_Stat - 2*math.log(TP)
	Fisher_Stat = Fisher_Stat - 2 * math.log(TP)


		Fisher_Stat = Fisher_Stat - 2*math.log(TP)

		Fisher_pval = 1 - chi2.cdf(Fisher_Stat, 2*L)

Update GIN #87

Are you sure you want to change the base?

Update GIN #87

Conversation

fengxie009 commented Dec 7, 2022

tofuwen left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

tofuwen commented Dec 7, 2022

fengxie009 commented Dec 17, 2022

tofuwen left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

tofuwen commented Dec 20, 2022

kunwuz commented Jan 31, 2023

fengxie009 commented Feb 3, 2023 via email

tofuwen commented Mar 22, 2023

kunwuz commented Oct 12, 2023