What To Work On?
OK, let's start by jotting down a list of ideas. Then we'll review the list and see what looks like a fun place to start.
Project Ideas
Choose Your Own Adventure Mobile App
I'm a big fan of simple text-based adventure games. When I was a kid I used to love reading Enid Blyton's "Famous Five" books. They were books where you would read a piece of the story and then have to choose how to proceed. Depending on your choice you would then turn to a different page number. If you made the right set of choices you would solve the case. Make the wrong choices, however, and the bad guys would catch you!
There are already a good number of apps out there that follow this paradigm. I have played and enjoyed some of the games published by Delight Games such as Detective's Choice and Zombie High.
I think this could be a good project for a few reasons. Firstly, the technological requirements of creating an app like this don't seem to be too high. The basic structure is to present some text and images and then render some text options for the reader to choose from. Based on what they choose, you then render some new text and images. Secondly, this project has the potential to use some fun (to me at least!) graph/tree structure and state management concepts. We'll need some way of tracking where the reader currently is in the story, what options they are permitted to choose from, and where each option should take the reader from there. If I want to get really ambitious then I can even start to picture a UI (User Interface) for authors to write their stories in. This project could be either very simple or pretty complex. This makes it a good candidate for having clear goals and deliverables but also perhaps some red flags for not letting scope creep get out of hand!
AI or Big Data - Stock Market Analytics
I'm a sucker for the stock market (and Poker!). There's just something about the idea of betting but trying to gain an edge using information, strategy and insight. One of my favorite hobbies is reading articles on Seeking Alpha where people write about their views on how various stocks will do as well as what trading strategies they think might work best. An idea that's slowly been germinating is to try and either automate or build a model to replace my regular reading of all these articles. When I really think about it, all I do each time I open one of these articles is go through the same series of logical steps. I glance at the author to see if it's someone I'm familiar with or if they have a high number of followers. This gives me a starting point for "is this person someone I should listen to"? Then I read the opening summary, which is a usually a few bullet points summarizing the gist of their article. Then, if it sounds interesting, I'll read the main body of the article, then scroll down to the comments section and see what people are saying. When reading the comments, I'm usually trying to gauge if most people agree with the author and/or if anyone has any strong arguments against their suggestions.
In this brave new world of AI and big data analytics, I wonder if it would be possible to write something to consume some of these articles, process them for general sentiment (e.g. buy Apple stock now!) and maybe also process the comments for general sentiment (e.g. this guy's an idiot, you should sell Apple!) and maybe output some kind of summary score (buying Apple today gets a score of 6/10, whatever that means!). If you could create something like that, then a fun next step might be to actually run a performance test against the stock market. A month ago, our model gave buying Apple a 9/10 (let's assume anything above a 5 means you should buy), did Apple outperform or underperform the base stock market index (e.g. DOW, FTSE or Nasdaq) in the last month?
I'm sure we're unlikely to stumble across a surefire way to "beat the market", but it sounds like fun and also a good project for playing with a lot of great technology. Off the top of my head this will need some data aggregation (API pulls or web scraping), some data scrubbing (clean up and structure the data into a useful format), some data modeling (dust off my amazing stats knowledge and brush up on the latest machine learning techniques, maybe even some natural language processing!) and perhaps some fun reporting infrastructure (charts, dials, maybe some alerts?).
AI/Big Data - Soccer Transfer Rumours
Another of my guilty pleasures is scouring the web for football (soccer) transfer rumours (rumors)*. I have lost too many hours of my life reading extremely low quality articles about whether player X was spotted at the airport on transfer deadline day, or their wife is out shopping for apartments in London because they're about to move. It's a very sad use of my time but the addiction is real people! If only there were some way to write something to automate reading all the gossip, assessing if any of it is worth paying attention to and then perhaps just sending me an alert when a particular probability threshold has been met. These days, there are even online betting websites where you can put money on if player X will move to a particular team. What if my amazing new algorithm can "beat the market"? Maybe the stock market has too many hedge funds packed full of analysts doing this kind of stuff for a hobbyist to beat them, but maybe football transfers is an untapped market? Probably not, but again, sounds like fun!
*British to American translation free of charge!
Automated Charity Giving
Project Management Tool
Next Steps
Project Ideas
Choose Your Own Adventure Mobile App
I'm a big fan of simple text-based adventure games. When I was a kid I used to love reading Enid Blyton's "Famous Five" books. They were books where you would read a piece of the story and then have to choose how to proceed. Depending on your choice you would then turn to a different page number. If you made the right set of choices you would solve the case. Make the wrong choices, however, and the bad guys would catch you!
There are already a good number of apps out there that follow this paradigm. I have played and enjoyed some of the games published by Delight Games such as Detective's Choice and Zombie High.
I think this could be a good project for a few reasons. Firstly, the technological requirements of creating an app like this don't seem to be too high. The basic structure is to present some text and images and then render some text options for the reader to choose from. Based on what they choose, you then render some new text and images. Secondly, this project has the potential to use some fun (to me at least!) graph/tree structure and state management concepts. We'll need some way of tracking where the reader currently is in the story, what options they are permitted to choose from, and where each option should take the reader from there. If I want to get really ambitious then I can even start to picture a UI (User Interface) for authors to write their stories in. This project could be either very simple or pretty complex. This makes it a good candidate for having clear goals and deliverables but also perhaps some red flags for not letting scope creep get out of hand!
AI or Big Data - Stock Market Analytics
I'm a sucker for the stock market (and Poker!). There's just something about the idea of betting but trying to gain an edge using information, strategy and insight. One of my favorite hobbies is reading articles on Seeking Alpha where people write about their views on how various stocks will do as well as what trading strategies they think might work best. An idea that's slowly been germinating is to try and either automate or build a model to replace my regular reading of all these articles. When I really think about it, all I do each time I open one of these articles is go through the same series of logical steps. I glance at the author to see if it's someone I'm familiar with or if they have a high number of followers. This gives me a starting point for "is this person someone I should listen to"? Then I read the opening summary, which is a usually a few bullet points summarizing the gist of their article. Then, if it sounds interesting, I'll read the main body of the article, then scroll down to the comments section and see what people are saying. When reading the comments, I'm usually trying to gauge if most people agree with the author and/or if anyone has any strong arguments against their suggestions.
In this brave new world of AI and big data analytics, I wonder if it would be possible to write something to consume some of these articles, process them for general sentiment (e.g. buy Apple stock now!) and maybe also process the comments for general sentiment (e.g. this guy's an idiot, you should sell Apple!) and maybe output some kind of summary score (buying Apple today gets a score of 6/10, whatever that means!). If you could create something like that, then a fun next step might be to actually run a performance test against the stock market. A month ago, our model gave buying Apple a 9/10 (let's assume anything above a 5 means you should buy), did Apple outperform or underperform the base stock market index (e.g. DOW, FTSE or Nasdaq) in the last month?
I'm sure we're unlikely to stumble across a surefire way to "beat the market", but it sounds like fun and also a good project for playing with a lot of great technology. Off the top of my head this will need some data aggregation (API pulls or web scraping), some data scrubbing (clean up and structure the data into a useful format), some data modeling (dust off my amazing stats knowledge and brush up on the latest machine learning techniques, maybe even some natural language processing!) and perhaps some fun reporting infrastructure (charts, dials, maybe some alerts?).
AI/Big Data - Soccer Transfer Rumours
Another of my guilty pleasures is scouring the web for football (soccer) transfer rumours (rumors)*. I have lost too many hours of my life reading extremely low quality articles about whether player X was spotted at the airport on transfer deadline day, or their wife is out shopping for apartments in London because they're about to move. It's a very sad use of my time but the addiction is real people! If only there were some way to write something to automate reading all the gossip, assessing if any of it is worth paying attention to and then perhaps just sending me an alert when a particular probability threshold has been met. These days, there are even online betting websites where you can put money on if player X will move to a particular team. What if my amazing new algorithm can "beat the market"? Maybe the stock market has too many hedge funds packed full of analysts doing this kind of stuff for a hobbyist to beat them, but maybe football transfers is an untapped market? Probably not, but again, sounds like fun!
*British to American translation free of charge!
Automated Charity Giving
Like all honorable citizens (of Earth, I'm technically only a permanent resident of the US), I care for my fellow man (and even wo-man!). Like most other honorable citizens I tend only to remember to give to charities when some big event happens. Ooh earthquake over there, here you go Red Cross, have some cash. The one exception is that about 15 years ago, I let a nice fellow from Oxfam talk me into signing up to donate some small amount (just a few pounds, I was a student at the time!) every month. Through a mix of generally liking the work that Oxfam does and also being way too lazy to actually change anything, I've let that small monthly donation stay active. If I actually run the numbers, I've probably now donated somewhere around 500 pounds (about $700) to Oxfam. That makes me practically a Rockefeller! Clearly automated charitable giving is where it's at.
What I think would be cool (and I'm sure things similar to this do exist) would be if there were a service where you could build and curate a portfolio of charities and causes that you hold dear. Then you could allocate some fixed amount, let's say $20 a month, that the service would then automatically draw from your account and split amongst the various charities (weighted as you see fit). Now, when some new cause becomes important to you, instead of doing a one time donation to something, you'd simply add it to your portfolio. It would start getting a share of the $20, or you might decide to bump up the $20 to $25 so as not to reduce the amount going to the existing charities in your pool. Now when it comes to tax time, you aren't checking your inbox for one off receipts from the Red Cross, you just log in to this service and download your year end statement of giving. I think there are some services that do something like this. This project would likely begin with doing some research into what options are already out there. If there is something close enough then the outcome of the project might be to just sign up for that service. If something like this doesn't really exist then this might be a fun project to take on. Potential pitfalls I see are dealing with money (can the money go direct from the donor to the charity so I don't have to be responsible for handling money?), is there a decent directory of charities for my users to choose from? How will I know where to send the money? Lots of things to figure out, but a very fulfilling project if it ended up becoming an actual service.
One project I've mused on for many years now is a better way to manage tasks and projects and lay out the sequence of steps it takes to complete something. At my prior job we even spent some time building in-house tools to help with project management and project specifications. This is one of those projects where I feel like I have a blurry vision of roughly what I would consider my dream project management tool to be but still can't quite put my finger on it. I've used several of the currently popular project tools such as Trello, Asana and Slack. They all have their own strengths but none of them seemed to do exactly what I wanted. If I do decide to work on this it might make sense to work on it as one of the first projects because if it actually pans out then I could use my new software to manage all the subsequent projects!
OK, I think that's enough ideas for now. Now I need to figure out which project to tackle first! There is some fun stuff in each idea. Part of me is tempted to do more than one project at a time but I really don't want to end up with 4 half completed projects and nothing concrete to show for it. I'm really tempted to start with the project management tool so that I have something to use to manage all the other projects. However, for my first project, I think I should choose just one thing and, in particular, make it a project with a clear path to completion, perhaps with most technology requirements being things I've worked with before. Looking back over the list, the project that comes closest to that would have to be the "Choose Your Own Adventure Mobile App".
So let's do it!
Comments
Post a Comment