bids-collaborative / purchasing Goto Github PK

This project aims to analyze the time-to-completion of POs, and determine if this is significantly different across people, departments, times of year, etc.

Project lead is @kaiweitan with assistance from @juanshishido

Edit:

Time to completion

Spend vs. quantity x unit price - spend is the amount paid so far, quantity * unit price is the total committed amount
Need some way to scale by size of department, # of personnel, etc.
Use this to find "Blanket POs" - get blanket PO details from Andrew
Come up with a list of blanket POs and their % complete
Keep brainstorming efficiency metrics, and then cluster departments together based off of this
Deliverable - Use the "time-to-completion" as a means of finding Blanket POs. Include a notebook that explains the analysis run so far, as well as the reasoning behind the particular analysis.

Hey guys, I hope you're all having a good break. I'll get to specific questions / issues with projects in a little while, but for now two quick things:

1. No meetings this week. Instead, we'll communicate via github / slack
2. We need some pretty pictures by this Monday for the short BIDS presentation I'm giving. This should be a good opportunity to get some input from people at BIDS and make connections.
3. I spoke with Aman and we're shooting for a brainstorm session in a few weeks...I'll keep you posted.

Post any questions / thoughts to this issue and we can discuss from there @kaiweitan @juanshishido @dariusmehri @nlin3330 @anthonysuen

This project will use classification methods to categorize / cluster POs based on what's in the description field, supplier name, etc.

Project lead is @juanshishido with assistance from @kaiweitan

Hey @berkeley-dsc/purchasing - this is a thread to coordinate information / images / etc for the BIDS final presentation. Here's the info:

Where - BIDS
When - May 6th, 6:00pm
What - Probably a 10-15 minute presentation with a bit more info than last time

Let me know if you guys can be there...this'll be the last official presentation for this project, so you should come and get some credit. There will also be a happy hour afterward.

I'm going to need some updated pretty pictures, and a few talking points for each project. We can discuss tomorrow.

It would be great if @agcinsf could be there as well...let me know if that's possible so I can tell Anthony etc.

Thanks for all your hard work guys, getting close to the finish line!

Do this here: https://github.com/berkeley-dsc/purchasing/wiki

Here we will discuss updates from each team meeting. These will be:

@juanshishido and me on Wednesday at 3:00pm in BIDS
@kaiweitan as a personal update
@nlin3330 and me on Thursday at 3:30pm in BIDS
@dariusmehri as a personal update

Please post any and all progress, questions, and comments here, and I will update after the meetings.

NOTES FROM MEETINGS

To Do
@me

1. send Nick Adams an e-mail about text processing-
2. figure out what "Service Order Request" is. If it means maintenance, see if it's seasonal at all.
3. Figure out what "coaches corporate credit cards" is
4. Spend vs. quantity*unit_price...why are these different?

@juanshishido

1. Come up with a hand-made word to category dictionary and see how this does
2. Create a list of most probable words
3. Use description along with manufacturer name
4. Come up with a list of categories we want to use
5. Split up word probabilities by department
6. Preprocess text data and remove unwanted words/characters
7. Create a word frequency matrix
8. Run LDA / clustering on this

@nlin3330 & @dariusmehri

1. Play with color coding of networks to make it more readable
2. Come up with an image / plot that shows something interesting for presentation on the Monday after spring break..
3. Start playing around with an analysis of average price per item @nlin3330

Note: Next week is spring break. I'll send out e-mails about meeting, but I'm assuming people may be gone. Be in touch via github so we can plan out what we'll present at the BIDS checkpoint meeting on the Monday after spring break.

Hey folks - In lieu of meetings this week, I'd like an update of where we are with respect to the current projects. For any of these, if you guys wanna meet to chat about things then I am happy to do so.

@nlin3330 and @dariusmehri - I see that you've pushed some commits to the repository recently. Is the code/notebooks in a format that would be worth taking a look at?

@juanshishido - Did you see the notes from the last brainstorm meeting? Should be some suggestions / goals for wrapping up in there. Let me know what progress you're making.

@kaiweitan - Was it clear what we were discussing at the meeting last Friday regarding blanket POs? Let me know if you'd like to talk about analyses / coding stuff to finish this.

Only a couple weeks left everybody, let's finish strong.

Hey all - I've been looking through the doodle poll to figure out when meeting times work for everybody. Now that we have sub-group projects, I wonder how you all feel about this:

We meet as an entire group every two weeks. This would be Thursdays at 3:30pm. This should be an hour-long meeting where we give updates to each project, go over admin stuff, and discuss challenges we're facing / ideas we've got / etc.

We meet in sub-groups every two weeks (on alternate weeks from full group meetings). These would also be an hour long, but focused on more nitty gritty details of the project. Times would be:
Wednesdays at 3:00pm for @kaiweitan @juanshishido
Thursdays at 3:30pm for @dariusmehri and @nlin3330

I will try to make it to all of these meetings, and hopefully @anthonysuen or @danielturek can make it to group meetings as well.

What do you guys think about this? Good idea? My concern about always meeting as a full group is that all of our schedules don't overlap, and we may not be able to cover questions in a particular sub-group to the extent that we'd like.

Let me know what you all think.

Get github user IDs, and add them to this team: https://github.com/orgs/berkeley-dsc/teams/purchasing

I've already given that team read/write access to this repo

TEST

It looks like this is going to be a fairly difficult thing to with with the current data. Here's what Andrew said:

Categorization is difficult. I can get you a “general classification” of the spend, but it will be messy.

He'll send something soon, but it looks like we may need to figure this out ourselves.

They should be able to sign up using their berkeley.edu address:

https://msdse.slack.com/messages/berkeley-dsc/

This issue is where we'll discuss the network analysis component. We can post graphs, code snippits, and brainstorms.

The network analysis project aims to find cluster of co-occurrence between departments, manufacturers, suppliers, product types, etc.

Project lead is @dariusmehri along with @nlin3330

This will be a short update meeting to discuss progress on the following projects:
#7 Network analysis
#8 Time to completio analysis
#9 Text classification analysis

Please post your progress here, as well as challenges that you've faced and things that need to be done next. As a group we can discuss these tomorrow.

@juanshishido @nlin3330 @dariusmehri @kaiweitan @anthonysuen

Chris

Hey guys, thanks again for sending me your work so far. In my (admittedly biased) opinion, this project was the most impressive of the ones that presented, you guys are great :)

We'll have a brief meeting tomorrow at 3:30pm. @anthonysuen, maybe you can meet to discuss the best course of action for the final stretch of this project?

Here's what I've got. Let me know if there's anything you'd like to discuss:

• Recap of what we've worked on so far. I'll go through the powerpoint presentation slides and you guys can give an update to the group of where we're at.
• Planning out the final month of this project. (we've got until the beginning of May so want to make sure there are some deliverables before then)
• Planning an analysis finalization session with Andrew + Aman from a data consulting group

@dariusmehri @anthonysuen @nlin3330 @dariusmehri @kaiweitan

Right now, we're focusing on the network analysis component to this project. There are many other ideas worth pursuing, and we should hash out some thoughts here. Here is the basic set of questions that Andrew asks when doing analyses:

1. What are we buying
2. Who are we buying it from
3. How much are we paying
4. Who on campus is buying it
5. How are we buying it

Additionally from Andrew, regarding what's useful regarding distributors vs. suppliers:

We go wherever there is most value. Anecdotally, our spend tends to be pretty fragmented at the manufacturer level. As a result, we focus on the distributors who are less fragmented. Ultimately, we can negotiate with either manufacturers or distributors and routinely ink contracts with both sets.

If you force me to pick… We typically have more leverage with distributors, simply because we can swap them out without impacting product choice on campus. When I tell someone they can’t have a specific product, they tend to get angry at me, especially as the complexity of the product increases.

That said, here is a preliminary list of ideas (italics means in progress):

1. Classification of the spend using a heuristic or machine learning. Maybe a logistic regression to predict the likelihood of being COTS Software? We have examples of COTS software and the ongoing maintenance transactions for them. Not a lot of examples, but probably enough.
2. Visualization of the processes used to buy. I suspect spend is traversing the procurement network in 10 or 20 different ways. Seeing those ways would show the scale of the problem and value of the opportunity.
3. Identifying the seasonality in the spend (if any). That would help identify when the need arises (generally) for software. I would guess there is a pattern to research buys that corresponds to the academic calendar, but have no idea. Or a forecasting in general would be cool.
4. Identifying (or predicting) the major places on campus (physical or departmental) where software buying takes place to identify the key markets. This would be cool as a bipartite graph showing relationships between departments on campus and software they buy. I.e. Matlab is used in Chemistry. Adobe Professional is used in Art History. Etc. Maybe using Gephi for some graph analysis?
5. Using text analysis and clustering/classification to assign "product types" to POs by looking through their description, supplier, etc.

Hey all - let's have a final work session to hash out final plots / graphs that we'll use in the presentation on Tuesday/Wednesday (or at least to make a clear road to create something by the weekend). We can give a brief update (just a few mins) and spend the rest of the time working to finish up what we've got.

Are you guys free for a working meeting around 3:00pm on Wednesday at BIDS? Let me know if this sounds good.

@berkeley-dsc/purchasing

Hey guys - last week we met individually, so tomorrow we've got a group meeting tomorrow to go over progress on everybody's project.

Primary agenda is:

• discuss what's left to be done on each project
• cover any questions that you guys have come across in the last few days.
• come up with a final format for deliverables

It looks like the final presentation for this project will be May 6th at 6:00pm at BIDS. I'll make an issue for this when I confirm that date.

Hey folks - just a reminder that we'll be meeting tomorrow at 2:00pm at BIDS. Let me know if you will not be able to be there in person.

Here's what we'll cover:

• Project updates and recap. I've got the presentation I gave at BIDS, but any new results are encouraged.
• Come up with goals for each sub-project. Think about results that would be useful for Andrew.
• Plan the documentation / wrapping up phase of this project.

Let me know if I'm missing anything, or if you want to make suggestions to the plan.

@berkeley-dsc/purchasing