seabbs
P2P Blogger
Posts: 41 
Likes: 22
|
Post by seabbs on Jun 26, 2017 9:59:29 GMT
Hi guys, I have been getting frustrated with the visualisation tools available for the funding circle loanbook - so I spent the weekend knocking up a quick shiny app. If your interested then check it out (note it is running on a small aws server so may be a little slow if it gets multiple users at once), and let me know if you found it useful. I am interested in developing it further, improving the UI, and adding additional tools. The next planned feature is to add the ability to upload your personal loanbook for analysis (this is something I am really interested in). If you fancy running it yourself locally then the code is provided on github, everything should work out of the box once the packages are installed, and the fc loanbook book is saved in the same folder as loanbook.csv. The GitHub repo has instructions in the readme if you haven't used R before. Links: www.seabbs.co.uk/shiny/fcdashboardwww.github.com/seabbs/fcdashboardTah |
|
|
|
Post by dan1 on Jun 26, 2017 11:08:20 GMT
Hi guys, I have been getting frustrated with the visualisation tools available for the funding circle loanbook - so I spent the weekend knocking up a quick shiny app. If your interested then check it out (note it is running on a small aws server so may be a little slow if it gets multiple users at once), and let me know if you found it useful. I am interested in developing it further, improving the UI, and adding additional tools. The next planned feature is to add the ability to upload your personal loanbook for analysis (this is something I am really interested in). If you fancy running it yourself locally then the code is provided on github, everything should work out of the box once the packages are installed, and the fc loanbook book is saved in the same folder as loanbook.csv. The GitHub repo has instructions in the readme if you haven't used R before. Links: www.seabbs.co.uk/shiny/fcdashboardwww.github.com/seabbs/fcdashboardTah Welcome to the forum seabbs. An impressive first post, thanks for sharing your work. I tried to take a look but it was running a little too slow to interrogate, a sign that it's popular. I never expected to see the use of R or principal components on this forum mind you. At least it wasn't written in SAS or SPSS!
|
|
seabbs
P2P Blogger
Posts: 41
Likes: 22
|
Post by seabbs on Jun 26, 2017 11:53:11 GMT
Thanks! Yes sorry about that - unfortunately it is hard to do anything about that as R/shiny is single process  . The other issue is when plotting scatters interactively it takes a while and is a bit dodge due to the number of data points. Haha yes - or an uncommented R script for that matter. Just as a note in case people aren't used to R - the plots are all interactive. This means that categorical variables can be selected (by double clicking on the legend), data ranges can be selected (by dragging), and plot values can be seen by hovering over them. You can also use the controls in the menu to select the ranges of dates to compare (for example excluding the first year of operation), select numeric variables to plot and select the categorical variable to break the data down by. |
|
|
|
Post by spiker on Jun 26, 2017 13:04:08 GMT
very nice
I know a little R, so this is always interesting!
|
|
seabbs
P2P Blogger
Posts: 41
Likes: 22
|
Post by seabbs on Jul 9, 2017 23:45:24 GMT
Hi Guys, Updates! Link: www.seabbs.co.uk/shiny/fcdashboardThe major reason I wanted to do this was to have a better understanding of the loans I was actually invested in - as the FC tools are very poor for looking at things at a macro level. Just updated the app with a personalised dashboard that enables rapid exploration of your personal holdings. If any of you find it useful and have feedback let me know. Updates in brief: - Added a personal dashboard for your loan book (requires data upload [data is stored only temporarily] or running the app locally) - Added exploratory plots of loans you are invested in - Added PCA of loans you are invested in - Updated loan book to 10/07/17 - Added ability to upload an updated copy of the loanbook - Updated the running a local version instructions. Futures updates: - Improved instructions - Comparison between FC and personal loan books - Summary measures attempting to capture the likely stochasticity of your returns. Any other suggestions welcome! |
|
seabbs
P2P Blogger
Posts: 41
Likes: 22
|
Post by seabbs on Jul 13, 2017 10:04:19 GMT
Quick update.
Link: seabbs.co.uk/shiny/fcdashboard
- Added facetting by variable (split your personal loanbook by a variable i.e loan term)
- Added filtering by variable (i.e look at only loans with a D risk)
- Updated interest rates to match annualised interest formula used by FC
Again any suggestions welcome!
|
|
seabbs
P2P Blogger
Posts: 41
Likes: 22
|
Post by seabbs on Jul 28, 2017 15:02:27 GMT
|
|
seabbs
P2P Blogger
Posts: 41
Likes: 22
|
Post by seabbs on Jul 28, 2017 16:38:04 GMT
|
|
ceejay
Posts: 971
Likes: 1,149
|
Post by ceejay on Jul 30, 2017 9:01:05 GMT
Hi
Just in case you're wondering if you're talking to yourself... I thought I'd let you know that I've had a go at running your package locally. FYI I tried and failed to get it to install fully on Linux (permission problems when I got to installing the tidyverse package) and TBH I didn't want to spend too much time finding the issue. So I installed it on a Windows machine and it seems to be running fine - though I am a complete novice with R and it will take me a while to see what I can really do with this.
Just one thing for now: I think you need to add "e1071" to the list of packages to be installed, otherwise the PCA options don't work
|
|
ceejay
Posts: 971
Likes: 1,149
|
Post by ceejay on Jul 30, 2017 9:15:34 GMT
And straight away my first bug: in the Personal Dashboard Overview the app is showing Amount Defaulted as zero ... which is sadly not true! Check that you are looking for "Bad Debt" in the "Loan status" column.
|
|
seabbs
P2P Blogger
Posts: 41
Likes: 22
|
Post by seabbs on Jul 30, 2017 11:24:17 GMT
Hi Ceejay,
Thanks for the feedback.
Yes, apologies for that - I don't have any bad debt so I had to guess how it was going to be encoded (I guessed Defaulted, oops).
Should now be fixed. I wasn't quite sure if it was encoded as Bad debt, Bad Debt but it is picking up both. Let me know if this hasn't fixed it for you!
In general, there may be a few things like this as obviously my personal loanbook is not representative of everyone else's, so just let me know if you find something.
I like using it to look at my loans by the number of repayments made (as this isn't something FC surfaces) - but I imagine that not everyone is so interested in this!
Any suggestions of what you would find helpful would be great.
|
|
seabbs
P2P Blogger
Posts: 41
Likes: 22
|
Post by seabbs on Jul 30, 2017 11:31:06 GMT
Sorry just saw your first comment as well.
Installing all the dependencies can be a pain with R - especially if it's new to you. I've added the package you mentioned to the list.
The permissions issue is not something I have ever run into - hmm how strange.
To make this easier for people I am setting up a docker container to run the app so watch this space!
|
|
ceejay
Posts: 971
Likes: 1,149
|
Post by ceejay on Jul 30, 2017 12:41:31 GMT
Thanks, yes I can confirm that buglet is now fixed.
I was interested by your comment about "payments made" as that is something that interests me too - and as you say, it's something you have to calculate from FC's data. But I don't see where you've made that available in this system. If I wanted to use the system to show me, for example, how many payments are made before a loan goes late or defaults, could you do that? I've knocked up something in Excel to do that but of course it's quite rough and you really want to see how this changes with volumes of loan, etc, which your system ought to be good at.
More generally, a few worked examples of settings to show different interesting plots would be very helpful.
|
|
seabbs
P2P Blogger
Posts: 41
Likes: 22
|
Post by seabbs on Jul 30, 2017 13:17:00 GMT
So it is currently only for your personal loanbook, it is one of the stratifying/filtering/faceting options. For example, if you filtered on loan status, selected only defaulted loans and then stratified by repayments made (see this example with dummy data: www.dropbox.com/s/ovxglu18rfm3kkn/Screenshot%202017-07-30%2014.13.01.png?dl=0). It's something I am planning on adding to the main loanbook as well - just need some time to get it sorted. I agree examples would be useful. I am waiting for it to be feature complete (ish) as they are quite time consuming to write and I only want to do it once!
|
|
seabbs
P2P Blogger
Posts: 41
Likes: 22
|
Post by seabbs on Jul 30, 2017 13:49:01 GMT
|
|
. The other issue is when plotting scatters interactively it takes a while and is a bit dodge due to the number of data points. 