Issues importing pytwoway

[SMaric93]

Hi, I'm having issues using pytwoway. I use Spyder as my IDE, although the same results are in Jupyter/Xcode, all running on lates MacOS. More specifically, when I try running "import pytwoway as tw" I get the following error:

runfile('/Users/PhD/Papers/AKM/Code/blm/code/untitled0.py', wdir='/Users/PhD/Papers/AKM/Code/blm/code') Traceback (most recent call last):

File ~/PhD/Papers/AKM/Code/blm/code/untitled0.py:9 in import pytwoway as tw

File ~/opt/anaconda3/envs/blm/lib/python3.9/site-packages/pytwoway/init.py:3 in from .twoway import TwoWay

File ~/opt/anaconda3/envs/blm/lib/python3.9/site-packages/pytwoway/twoway.py:8 in class TwoWay():

File ~/opt/anaconda3/envs/blm/lib/python3.9/site-packages/pytwoway/twoway.py:104 in TwoWay def cluster(self, measures=bpd.measures.cdfs(), grouping=bpd.grouping.kmeans(), stayers_movers=None, t=None, weighted=True, dropna=False):

AttributeError: module 'bipartitepandas.measures.measures' has no attribute 'cdfs'

I appreciate any help! Thanks!

[adamoppenheimer]

I apologize for this issue.

This is because I have been working on some major changes to pytwoway and the code uploaded to pip, as well as the documentation, are both out of date.

Here is how to get the code working for now:

• First, make sure you have bipartitepandas installed and updated by running pip install bipartitepandas if you don't already have it installed, and pip install --upgrade bipartitepandas if you already have it installed.
• Next, you must clone the repository for pytwoway locally. This can be done by going to the main page, clicking on the green code link, and downloading then unzipping the file wherever you would like it to be located.

The following is some code that should get the code running for the BLM estimator (please let me know if it's not working for you):

# Add PyTwoWay to system path (SET THIS TO WHERE YOU HAVE THE PACKAGE REPOSITORY CLONED)
import sys
sys.path.append('../../..')

###########################
##### Import packages #####
###########################

import numpy as np
import bipartitepandas as bpd
import pytwoway as tw
from matplotlib import pyplot as plt

#######################################
##### Check out parameter options #####
#######################################

print('----- BLM -----')
tw.blm_params().describe_all()
print('----- Clustering -----')
bpd.cluster_params().describe_all()
print('----- Cleaning -----')
bpd.clean_params().describe_all()
print('----- Simulation -----')
bpd.sim_params().describe_all()

#################################
##### Set parameter choices #####
#################################

nl = 2 # Number of worker types
nk = 3 # Number of firm types
blm_params = tw.blm_params({'nl': nl, 'nk': nk})
cluster_params = bpd.cluster_params({'measures': bpd.measures.CDFs(), 'grouping': bpd.grouping.KMeans(n_clusters=nk), 'is_sorted': True, 'copy': False})
clean_params = bpd.clean_params({'drop_returns': 'returners', 'copy': False})
sim_params = bpd.sim_params({'nl': nl, 'nk': nk})

#########################
##### Simulate data #####
#########################

sim_data = bpd.BipartiteDataFrame(bpd.SimBipartite(sim_params).simulate()[['i', 'j', 'y', 't', 'l']]).clean(clean_params).cluster(cluster_params).collapse(is_sorted=True, copy=False).to_eventstudy(is_sorted=True, copy=False)
print('Movers data')
display(sim_data[sim_data['m'] > 0])
print('Stayers data')
display(sim_data[sim_data['m'] == 0])

###########################################
##### Initialize and run BLMEstimator #####
###########################################

# Initialize BLM estimator
blm_fit = tw.BLMEstimator(blm_params)
# Fit BLM estimator
blm_fit.fit(jdata=sim_data[sim_data['m'] > 0], sdata=sim_data[sim_data['m'] == 0], n_init=40, n_best=5, ncore=1)

###############################
##### Investigate results #####
###############################

# Plot likelihood vs. connectedness
blm_fit.plot_liks_connectedness()

# Plot estimated worker-firm pair means for first period
blm_fit.plot_A1()

[SMaric93]

Hi Adam, It works, thanks so much! Stefan