mood & emotion
About three months ago, Vik Parachuri was in a situation where he thought he was unhappy. He was working a lot and didn't know how to fix his mood. Wanting to measure his happiness and mood to improve it, he tried a lot of measurement tools, android apps, and web pages, but most of them did not meet his needs. He then went on to create an app called Happsee, which collects data using smartphone sensors to quantify and track happiness. In this talk, he discusses the app and what he has learned from it.
Happsee | phone
Basically three months ago I was in a situation where I thought I was unhappy. So I thought I was in this situation where I thought I was unhappy, I was working a lot. I didn’t know if I was unhappy, one, and if I was unhappy, two, I didn’t know how to fix that. So I wanted to explore this whole space of unhappiness and how to improve it.
So how do we do that? I wanted to measure my happiness and mood. I think one way to do that is to track passive data. You can’t only do that from passive data though and I wanted to ways that would enter things in and correlate with passive data and do all sorts of cool things like machine learning.
So what can we use? So I looked at a lot of measurement tools. I looked at a lot of android apps. I looked at a lot of web apps. Most of them did not suit my needs. They were either too complex. They asked me to fill out a bunch of surveys and I think a few of them actually made me angry, because I was like why am I filling out all these questions and that completely defeats the purpose of understanding whether I’m happy. And some of them were really trivial, like slide up or down and make this happy face a frowny face if you’re sad and that seemed kind of ridiculous to me to.
So in the spirit of software I made my own tool. I made this android app called Happsee. Basically my goal was to maximize information gathered while minimizing the information I actually hat to type in because I’m lazy and why not. And I also wanted to get periodic reminders, so I wanted it to tell me to do things.
So here’s some screenshots of that toll. This is the main entry. So I wanted to minimize the information I entered. So I entered in some moods. I entered in how happy I was on a zero to ten scale. I could type some stuff in if I wanted and I could attach a photo. And using that it could generate a bunch of graphs. It could do some cool analysis and tell me different this like when your engaged your this happy, show me my photos, etc.
So I’ve been tracking my happiness since October. Actually this chart is since November 1 2013. So you can see I did not have a good December and I guess I didn’t really like my holidays, but now I’m much better, I’m good. You got me at my peak. Aren’t you happy guys? So this is me at my peak. It doesn’t get any better than this.
And here’s a cool happiness map. So as you can see I live around here on the big yellow spot. So sometimes I’m happy when I’m home, sometimes I’m not I guess and that averages out. I tend to be happier when I’m outside, and the power of this cool map is that you can zoom in and out.
So I guess I like Boston more than New York and I have some dots in New York, so if any of you are New Yorkers. I like Kansas but I’m not going to zoom out far enough to get Kansas.
So here are my moods over time. These are my negative moods specifically. So we can see that these have changed. So basically what this is is the top means that I’m really feeling that mood over time, and this is a smoothed average. And the units are kind of what you saw when you entered the moods there was a binary zero one thing, and since this is a smooth average I did a zero to ten scale. So a ten means I’m basically feeling the mood like all the time, a zero means that I’m not feeling it at all.
So this is tired this top line. So I was really tired and now I’m not so tired, but I’m happier now so maybe those two have something to do with each other. And you can see the other ones are changing here to. So uncertain is dropping. Stressed was in don’t know, I was never really that stressed, and bored is dropping, so I’m not bored at all now. So negative moods have been changing and my positive moods have also been changing and they’ve also been going up. So I’m more energetic which is this line down here, I’m more relaxed, more engaged and I am also more confident I guess maybe.
And this is the cool thing. So recently I added the ability to this app to track passive factors, and I don’t have any cool charts like correlating these passive factors to different moods. But basically what you’re seeing here is a chart when I added this functionality which was I think a the beginning of this month till now.
So tilting this top one is when I’m using my phone. This is when the phone is still, when I’m not carrying it, when I’m not doing anything with it. this is when I’m walking around and this is when I’m in a vehicle. So I’m almost never driving. Sometimes I walk around I guess, most of the time I’m using my phone, sometimes I’m sitting around.
So this is a way to track that activity over time, and you can imagine getting some really cool correlations with this data like for example, when you use your device, you’re unhappy, maybe you don’t use your phone so much.
And this is just tracking periodic locations. So the blue dots are the more recent ones, so that’s where I was more recently, and the grey dots are further back in time so this is basically tracking my location passively every hour or so. And there’s a couple of like other passive things that I’m tracking, and I can talk about that in the questions if you want.
So this is a cool word cloud. Basically these are a bunch of words that I mention in my entries, and the greener something is the happier I am when I mention that, and the less green, the more yellow the more unhappy I am.
So being sick makes me unhappy, go figure, green and really green make me happy. Priya is actually my girlfriends name and she was like, you like meetings more than me. I was like yeah I guess, meetings are my thing.
But basically I think the key takeaway here is that people make me really happy, being out with my girlfriend, and I really like Happsee apparently for some reason.
Kansas was when I went with my girlfriend’s family, and like had Thanksgiving with them and stuff and I apparently liked that, so it’s kind of interesting.
And I was having a lot of sleep issues, so mentioning sleep made me really unhappy because I wasn’t getting good sleep. And being in new circumstances generally makes me happy. I repeated this slide because of the Ignite format and sometimes you don’t get a chance to say everything that you want, and it auto advances the slide. Only one person is nodding so I’ll move on from that point. But anyways this is a duplicated slide.
Okay, can you guess what these correspond to? So this top one here is tired, this is happiness and this is energy. So as you can see being tired had a huge effect on my happiness, and when I started to get les tired, I started to get a lot more happy. And having this data basically let me start making modifications to how I slept, like using black-out curtains, going to sleep at a certain time, and that really led to me being a lot less tired and it really improved my happiness.
And this is why I created this tool, like things like this are insights that maybe I would have had otherwise, but I probably wouldn’t have and it’s like this huge insight. And you look back and you’re like well maybe that’s not so big but it can change your life. So obviously it is big.
So little insights like this, and I’ve also realised again that getting out and talking to people makes me a lot happier, so I’m doing a lot more of that and giving presentations like this and getting outside of my comfort zone, and these are all think I like that I didn’t know about until I started measuring them.
So what am I thinking for the future? So I want to use machine learning to predict happiness. So I’m lazy, and the one thing that gets me annoyed a little bit is sometimes I have to make a lot of entries to tell the system how happy I am. What if it could just pick that up from passive data and I think that’s one thing I want to explore more.
Apply my own experiences from the past to the future. As Daniel Gilbert, the psychologist said, everyone has already experience something that someone else has experienced and you’ve probably experienced things that you’ve experienced in the past. So I would love to translate that from my past to my future, and say, I lost my job a year ago and I got happier in a week. So if I lose my job again it probably won’t have such a big impact on me as I expect.
Do image analysis to figure out emotion, an if I’m always taking pictures of other people and I’m always taking pictures of sunlight, maybe there is a correlation of happiness of maybe what I’m taking pictures of has a correlation with my happiness.
Okay, so again the experience sharing shared understanding, I think it’s really important to connect to the people around you. Happiness isn’t just a thing about you. it’s about how you network with the people around you and how you interact, and there’s definitely a lot to explore there.
Pulling data from other sources, I’ve started to track some passive stuff, but you can do things like integrate with like RunKeeper and integrate with Fitbit, Yelp, whatever you want to do, all kinds of API’s to play with.
So if you want to try it it’s in beta for Android. You can go to the cool website or email me, whatever floats your boat. I have a Twitter to but it’s not on here.
Anyways that’s it questions.