p-p-p-p-p-p-p-p-p-p- wrote:

We're not training an AI model or anything, it's just going to be an algorithm with a little bit of randomness. Basically the idea is you submit your constraints (for example, I only want to train this muscle group, or I only have dumbells) and it filters all the workouts that meet those constraints, takes pre assigned “popularity” values for those workouts, and enter them into a pool and pick randomly with a higher likely hood for the more popular workouts.

isnt that what a search engine does with the query being their preferences?
Just that it picks only the top 5 to show

p-p-p-p-p-p-p-p-p-p- wrote:

Talking about popularity, I think we should make a project that helps us determine the popularity of workouts. If I made the basis of a project, would you be able to set up a database to connect it?

I think so yea but to find out the popularity of a workout or exercise
we need popularity on the actual project to get this info

p-p-p-p-p-p-p-p-p-p- wrote:

Basically, the project would give users 25 random workouts and ask them how familiar they are with the workout, and then give or take away “popularity” from that workout based on the users other answers (if a user give a lot of “unfamiliar” ratings, the “familiar” ratings are worth more to that workouts popularity ranking, but if the user gives a lot of “familiar” ratings, the 'unfamiliar“ rankings are worth more to that workouts popularity ranking.) Obviously that sounds like a lot, but I think I can easily make a project without a backend in an hour or two. So if you can get a backend up by tonight, and we properly advertise the project, then we can get a lot of valuable data. Idk if that would count as ”training" data or not, but it seems like it might be!

Well for that it is getting late but i can try to make it as i dont expect it to take too long like more than an hour or two

and as for the exercises wont it be hard to select for each one if they are familiar
How about we just wait and record which workouts/exercises are selected the most
that way the user wont have to select anything
But we will need to wait and gather a lot of information

as for how to call it i dont think it really matters that much I did some brief research and found out its called targeted advertising
"Targeted advertising is all about collecting and analyzing data about a user, or preferably a group of users."
But we can call it whatever we want

p-p-p-p-p-p-p-p-p-p- wrote:

Let me know if you are confused about any of this.

I think i understand but do have some questions

p-p-p-p-p-p-p-p-p-p- wrote:

As for workout codes, I think they're compressed enough. The “circuit” workouts are just workouts where you're doing 2-3 different workouts over and over again, so those can easily be compressed. There's really 3 things that are important to this code, though:

1. The ability for the system to recognize a pattern, if there is one, and convert the code into a circuit on behalf of the user.
2. A new GUI for the # of the circuit you're on, this is going to be a small icon but it might be problematic if we don't have room for it
3. a way to decode the code into the system so we can use pre-existing code readers instead of having to make a new one.

2. Yea it is a bit challenging we do need to free up some space design wise

Oh ok anyways going to start making the simple database now

So just a brainfart here, but this is how the Workout Survey is going to work:

1. All workouts start at 0 “points”
2. The user answers questions about 20 random workouts. Once this occurs, their data is then submitted to a database which processes it (until we get the database, users will submit their data in a forum post, and I will use a decoder to import it into the system)
3. The way the users data works is initially, every “I know” is worth +5, every “I don't know” is worth -5, and every “I've heard of it but I don't fully know what it is” is worth 0. However, once they've answered about all 20 workouts, their answers will be re-distributed so that if they answered more one way, it's worth less (for example, if a user doesn't know about a lot, then those answers shouldn't have as much as an impact, but the things that they do know will be worth more. However, if a user knows a lot, then what they DO know isn't worth as much to us, but what they DONT know is worth a lot more.)
3.5: To balance this, the project will also keep track of which workouts haven't gotten that much attention, so they're about even when the survey period is over.
4. Once we have collected sufficient data (lets say 30-50 answers for each workout) we will adjust all the values so that the highest one is 100 and the lowest is 0. This way, we don't have a large variety of numbers and they're easy to work with.

p-p-p-p-p-p-p-p-p-p- wrote:

So just a brainfart here, but this is how the Workout Survey is going to work:

1. All workouts start at 0 “points”
2. The user answers questions about 20 random workouts. Once this occurs, their data is then submitted to a database which processes it (until we get the database, users will submit their data in a forum post, and I will use a decoder to import it into the system)
3. The way the users data works is initially, every “I know” is worth +5, every “I don't know” is worth -5, and every “I've heard of it but I don't fully know what it is” is worth 0. However, once they've answered about all 20 workouts, their answers will be re-distributed so that if they answered more one way, it's worth less (for example, if a user doesn't know about a lot, then those answers shouldn't have as much as an impact, but the things that they do know will be worth more. However, if a user knows a lot, then what they DO know isn't worth as much to us, but what they DONT know is worth a lot more.)
3.5: To balance this, the project will also keep track of which workouts haven't gotten that much attention, so they're about even when the survey period is over.
4. Once we have collected sufficient data (lets say 30-50 answers for each workout) we will adjust all the values so that the highest one is 100 and the lowest is 0. This way, we don't have a large variety of numbers and they're easy to work with.

That seems about right
But would anyone want to do the survey
i guess its not a bad idea while im still working on the cloud database we could do surveys but after that we can record which exercise gets used more often
and implement a point balance system like you said above so we can be getting data when someone just uses SweatFit

NIKI-KOLCHAGOV wrote:

p-p-p-p-p-p-p-p-p-p- wrote:

So just a brainfart here, but this is how the Workout Survey is going to work:

1. All workouts start at 0 “points”
2. The user answers questions about 20 random workouts. Once this occurs, their data is then submitted to a database which processes it (until we get the database, users will submit their data in a forum post, and I will use a decoder to import it into the system)
3. The way the users data works is initially, every “I know” is worth +5, every “I don't know” is worth -5, and every “I've heard of it but I don't fully know what it is” is worth 0. However, once they've answered about all 20 workouts, their answers will be re-distributed so that if they answered more one way, it's worth less (for example, if a user doesn't know about a lot, then those answers shouldn't have as much as an impact, but the things that they do know will be worth more. However, if a user knows a lot, then what they DO know isn't worth as much to us, but what they DONT know is worth a lot more.)
3.5: To balance this, the project will also keep track of which workouts haven't gotten that much attention, so they're about even when the survey period is over.
4. Once we have collected sufficient data (lets say 30-50 answers for each workout) we will adjust all the values so that the highest one is 100 and the lowest is 0. This way, we don't have a large variety of numbers and they're easy to work with.

That seems about right
But would anyone want to do the survey
i guess its not a bad idea while im still working on the cloud database we could do surveys but after that we can record which exercise gets used more often
and implement a point balance system like you said above so we can be getting data when someone just uses SweatFit

Its getting quite late so i think im going to continue this tomorrow

Well I want to polish the AI this week, so I'd prefer if we get this data ASAP. But you're right, while people are making workouts we can potentially collect data (if we can get a database set up) in order to continually update the AI and keep it relevant.

p-p-p-p-p-p-p-p-p-p- wrote:

Well I want to polish the AI this week, so I'd prefer if we get this data ASAP. But you're right, while people are making workouts we can potentially collect data (if we can get a database set up) in order to continually update the AI and keep it relevant.

Yea well its tomorrow now or today im going to continue making it hopefully i can get it working

I'm going to pause this right now so I can spend time with my grandparents, but I'll be back in 24 hours (maybe a little bit more)

p-p-p-p-p-p-p-p-p-p- wrote:

I'm going to pause this right now so I can spend time with my grandparents, but I'll be back in 24 hours (maybe a little bit more)

oh ok

p-p-p-p-p-p-p-p-p-p- wrote:

I'm going to pause this right now so I can spend time with my grandparents, but I'll be back in 24 hours (maybe a little bit more)

same, I'm going this weekend abroad

Alright I’m going to be busy today too, but I’m going to try to finish off the survey tonight

My school starts in 1 month and two days.

Until then, I am going to put a lot of my energy into this project to get it as shaped up as possible for a potential 2.0 release in late november (exactly 1 year after the 1.0 release, so 2.0’s update cycle directly correlates with that of 1.0’s update cycle).

However, once school starts, I don’t know how much time I will have on my plate, especially considering the sheer amount of commitments I currently have.

That means we need an all hands on deck. Some of you guys are working on side projects for SweatFit, which sound really cool, but I would prefer if those make a later release. All the SweatFit resources we have need to go into 2.0 right now.

Ok, I understand. Just one question though. Any art or anything needed? I'm all in to help!

p-p-p-p-p-p-p-p-p-p- wrote:

My school starts in 1 month and two days.

Until then, I am going to put a lot of my energy into this project to get it as shaped up as possible for a potential 2.0 release in late november (exactly 1 year after the 1.0 release, so 2.0’s update cycle directly correlates with that of 1.0’s update cycle).

However, once school starts, I don’t know how much time I will have on my plate, especially considering the sheer amount of commitments I currently have.

That means we need an all hands on deck. Some of you guys are working on side projects for SweatFit, which sound really cool, but I would prefer if those make a later release. All the SweatFit resources we have need to go into 2.0 right now.

Oh Ok

For me school starts after about 1 month as well
But while i am probably going to be busy i think i would have some time between 6pm - 1am
To work on version 2.0

NIKI-KOLCHAGOV wrote:

p-p-p-p-p-p-p-p-p-p- wrote:

My school starts in 1 month and two days.

Until then, I am going to put a lot of my energy into this project to get it as shaped up as possible for a potential 2.0 release in late november (exactly 1 year after the 1.0 release, so 2.0’s update cycle directly correlates with that of 1.0’s update cycle).

However, once school starts, I don’t know how much time I will have on my plate, especially considering the sheer amount of commitments I currently have.

That means we need an all hands on deck. Some of you guys are working on side projects for SweatFit, which sound really cool, but I would prefer if those make a later release. All the SweatFit resources we have need to go into 2.0 right now.

Oh Ok

For me school starts after about 1 month as well
But while i am probably going to be busy i think i would have some time between 6pm - 1am
To work on version 2.0

Starting from tomorrow I have to catch a plane for two weeks

p-p-p-p-p-p-p-p-p-p- wrote:

Well I want to polish the AI this week, so I'd prefer if we get this data ASAP. But you're right, while people are making workouts we can potentially collect data (if we can get a database set up) in order to continually update the AI and keep it relevant.

Time for PHP and SQL…

Here is the new UI I was talking about.

I need Niki to create a NewUI Concept for it so I can implement it. You have most artistic freedoms with it as long as it conveys the same information, and is a popup menu. This was just the way I would implement it.

Circuit is 100% complete, and it will even detect workouts that may have circuits within them. That was a very complicated algorithm.

Also, the project now has over 7,150 blocks of code! Woohoo!

The next item on the agenda is an everything menu, which hopefully solves the problem Niki brought up which is that there are too many icons on the screen. I'll release screenshots of it when it's in development.

Also Niki, for both of these things, if you are not overwhelmed by the stuff I have already talked about (which it's totally okay if you are, just let me know) I would like if you could dub over it with the new UI. both of these should be easy things to dub over, so you don't have to worry much about it.

Sick and tired of all the clutter? Welcome the Everything Menu, a menu that compiles all existing settings/customization features into one sleek menu.


Early design, icons will change over time.

Note: This is DIFFERENT than all the other things I've been talking about. I had this idea today. I still am working on the workout menu. Two seperate things

p-p-p-p-p-p-p-p-p-p- wrote:

Here is the new UI I was talking about.

I need Niki to create a NewUI Concept for it so I can implement it. You have most artistic freedoms with it as long as it conveys the same information, and is a popup menu. This was just the way I would implement it.

Oh Ok Looking great just that i dont entirely understand what is it about
Nevermind took me a while to get it but now i understand what you are trying to show

As for the designs i can start them now
But we need an icon for each exercise which might take quite a bit (not a priority as of right now)

p-p-p-p-p-p-p-p-p-p- wrote:

Circuit is 100% complete, and it will even detect workouts that may have circuits within them. That was a very complicated algorithm.

Also, the project now has over 7,150 blocks of code! Woohoo!

The next item on the agenda is an everything menu, which hopefully solves the problem Niki brought up which is that there are too many icons on the screen. I'll release screenshots of it when it's in development.

Also Niki, for both of these things, if you are not overwhelmed by the stuff I have already talked about (which it's totally okay if you are, just let me know) I would like if you could dub over it with the new UI. both of these should be easy things to dub over, so you don't have to worry much about it.

May i get some more information on circuit workouts they kind of look like sets but at the same time different

Shouldnt take to long to design these UIs
As i have most things done already but it will probably take some time
Will try to get them as early as i can

Also i have some stuff to finish the cloud database