Irish Poker Boards
Register Arcade FAQ Members List Calendar Mark Forums Read

Go Back   Irish Poker Boards > Community, Tech, and Feedback > Personal Logs
Reply
 
Thread Tools Display Modes
Old 05-09-17, 10:15   #21
Hitchhiker's Guide To...
Duncan F*****g Stewart
 
Hitchhiker's Guide To...'s Avatar
 
Join Date: Jan 2010
Location: South Cidre
Posts: 28,764
Quote:
Originally Posted by Hitchhiker's Guide To... View Post
Step 1.7: Complete the Data Scientist with Python program on DataCamp.

Read a lot of great things about this. The general view is that its not particularly great for learning from scratch, but if you have a few relevant uni courses under your belt then its excellent for getting the training in and getting you on top of the technicalities.

This is going to take a while. Have only a few spare evenings in August, and September is the start of intensive teaching. The plan though is to get it finished by the end of September.

Also going to pick up this hotly awaited book when the second edition is released at the end of September. Python for Data Analysis. The current edition is pretty much the pandas bible, but its now five years old so way too dated.

After finishing this I should be at an upper-intermediate level in general python and pandas, and maybe somewhere in the middle of the beginning of machine learning.
Skipped this for a bit as August was stupidly busy, and instead did two courses on Udemy on Tableau, the data visualisation software. You can usually pick Udemy courses up for a tenner when they have one of their regular sales. Did an introductory Tableau 10 A-Z and a more advanced course: Tableau 10 Advanced Training.

These courses are excellently delivered as the training is centred around case studies and Tableau is a phenomenal software. Almost to the extent that you would think anyone still using Excel to get information from data is a simpleton of the highest order.

Will move on to the planned DataCamp course next but it will be slow moving due to having to do some proper work.
Hitchhiker's Guide To... is offline   Reply With Quote
Old 05-09-17, 10:18   #22
Hitchhiker's Guide To...
Duncan F*****g Stewart
 
Hitchhiker's Guide To...'s Avatar
 
Join Date: Jan 2010
Location: South Cidre
Posts: 28,764
Quote:
Originally Posted by Hitchhiker's Guide To... View Post
After that then it'll be jumping into the famous Stanford Machine Learning course taught by Andrew Ng. Hadn't done this course up to this point as its in MATLAB which is a whole new software to get on top of.

The course strength seems to be in explaining why we're doing all these techniques and how we can choose between the various testing choices.

Then it'll be just getting more and more advanced in the area. At a rough guess then it would be to be at an upper-intermediate level in machine learning by the end of 2017, and some level of decent expertise by the end of the first half of 2018.
Since posting this, Andrew Ng has launched a new four-part course on deep learning set in Python that has the machine learning world all a flutter. Looks excellent. Might do this instead of his standalone course given its fairly clear from the trends that Python is going to rule the world for machine learning even if R is currently the best overall package.
Hitchhiker's Guide To... is offline   Reply With Quote
Old 01-10-17, 20:47   #23
Hitchhiker's Guide To...
Duncan F*****g Stewart
 
Hitchhiker's Guide To...'s Avatar
 
Join Date: Jan 2010
Location: South Cidre
Posts: 28,764
Making progress on the DataCamp 'Data Scientist with Python' track. Its 20 modules, and have six done in September, so will be finished at this rate around mid-November.

DataCamp have come up with something very smart. Very nicely delivered and designed. My path to date was the two intro uni of Toronto python courses followed by three applied data science courses from Michigan U. I'd highly recommend slotting this in the middle between Toronto and Michigan, as Michigan was far too advanced compared to the beginning skillset you would have at this stage.
Hitchhiker's Guide To... is offline   Reply With Quote
Thanks From:
Old 01-10-17, 20:50   #24
Hitchhiker's Guide To...
Duncan F*****g Stewart
 
Hitchhiker's Guide To...'s Avatar
 
Join Date: Jan 2010
Location: South Cidre
Posts: 28,764
In not quite related news also read two machine learning general books:

The Master Algorithm (how machine learning will save us all)

Weapons of Math Destruction (how machine learning will destroy us all)

Nice to have that bit of a broader perspective on what all that means, but The Master Algorithm despite being interesting, went into way too much detail on the technicalities of the various machine learning choices.
Hitchhiker's Guide To... is offline   Reply With Quote
Old 01-10-17, 20:56   #25
Hitchhiker's Guide To...
Duncan F*****g Stewart
 
Hitchhiker's Guide To...'s Avatar
 
Join Date: Jan 2010
Location: South Cidre
Posts: 28,764
Going to round out the year with a small dive into Excel VBA (got this course on Udemy for a tenner yesterday) and then a deepish dive in R.

My thinking is just to be familiar enough with all the main softwares that are used for some sort of data analysis. Excel VBA is still shockingly popular especially in Europe, so spending 4 hours getting up to speed on the main essence of it seems like a small thing. While R is highly popular in finance coding, even if trends suggest that Python will become the one language to rule them all. Already have Tableau (as noted above), not that thats anything to do with coding; and Stata from my research. So can feel quite assured then that I'm pretty much up to speed on just about every type of base software for data science. (Matlab seems to have been ditched by nearly everyone except academia).
Hitchhiker's Guide To... is offline   Reply With Quote
Old 04-11-17, 00:59   #26
Hitchhiker's Guide To...
Duncan F*****g Stewart
 
Hitchhiker's Guide To...'s Avatar
 
Join Date: Jan 2010
Location: South Cidre
Posts: 28,764
Quote:
Originally Posted by Hitchhiker's Guide To... View Post
Making progress on the DataCamp 'Data Scientist with Python' track. Its 20 modules, and have six done in September, so will be finished at this rate around mid-November.

DataCamp have come up with something very smart. Very nicely delivered and designed. My path to date was the two intro uni of Toronto python courses followed by three applied data science courses from Michigan U. I'd highly recommend slotting this in the middle between Toronto and Michigan, as Michigan was far too advanced compared to the beginning skillset you would have at this stage.
Ugh. There will be nothing done on this until December. Working so many hours just to stand still at the moment and doing the French driving theory test has eaten up nearly every free hour. Was delaying it week-to-week but decided today it was best just to defer even thinking about doing more courses until December. Then it's nine glorious months of freedom and can hopefully become expert in everything machine learning.
Hitchhiker's Guide To... is offline   Reply With Quote
Thanks From:
Old 05-01-18, 08:35   #27
Hitchhiker's Guide To...
Duncan F*****g Stewart
 
Hitchhiker's Guide To...'s Avatar
 
Join Date: Jan 2010
Location: South Cidre
Posts: 28,764
So pretty much nothing done on this September-December. Just finished this short Excel VBA course in order to have a flavour of the language.

Kicking it all off again properly starting next week. Quite content with the 2017 progress though.
Hitchhiker's Guide To... is offline   Reply With Quote
Old 05-01-18, 09:42   #28
brady23
Member
 
Join Date: Apr 2010
Posts: 2,210
Quote:
Originally Posted by Hitchhiker's Guide To... View Post
So pretty much nothing done on this September-December. Just finished this short Excel VBA course in order to have a flavour of the language.

Kicking it all off again properly starting next week. Quite content with the 2017 progress though.
That's good timing.

I literally popped in to check whether you recommended any VBA course as I need it for the modelling class and I have no experience with it.
I was told that because I can write SQL I should pick it up relatively quick but I'm not sure how true that is.
brady23 is offline   Reply With Quote
Old 05-01-18, 10:20   #29
Hitchhiker's Guide To...
Duncan F*****g Stewart
 
Hitchhiker's Guide To...'s Avatar
 
Join Date: Jan 2010
Location: South Cidre
Posts: 28,764
Quote:
Originally Posted by brady23 View Post
That's good timing.

I literally popped in to check whether you recommended any VBA course as I need it for the modelling class and I have no experience with it.
I was told that because I can write SQL I should pick it up relatively quick but I'm not sure how true that is.
It's actually very straightforward. That course gives all the basics, but you would need something extra to actually apply it in a finance context. A surprisingly nice language, given its Microsoft.
Hitchhiker's Guide To... is offline   Reply With Quote
Old 30-01-18, 23:42   #30
Hitchhiker's Guide To...
Duncan F*****g Stewart
 
Hitchhiker's Guide To...'s Avatar
 
Join Date: Jan 2010
Location: South Cidre
Posts: 28,764
So a lot of progress on this.

1. Done about five more courses on DataCamp specialism, but it's speeding up rapidly, and the rest of the courses in their python for data science specialism will be done in next two weeks.

2. Invited to write a book chapter on potential for machine learning in fintech. That should add a nice bit of public perception of expertise, as it's going to be a very well received book.

3. Invited myself, and the editor accepted, to write a journal article in a very good academic journal on potential for machine learning in finance research.

4. There's a team in work out fundraising for a chair I'll run in machine learning and finance (chair is the French word for research centre). Think that'll get in a fair bit of cash to fund diligent PhDs and postdocs.

5. Proposed and got approved a new python module for finance masters in work, that means all my teaching next year will be on machine learning applications to finance. That's first module of its type in France as far as can see.

So this is all quite positive. Essentially by end of year I'll have a very strong public identity as a machine learning expert in finance, with CV to match.

The issue then becomes how to translate that into cash. A big move up in uni rankings would be the logical step as every business school will soon be realising they have to have an expert in this area, and there's almost no-one they can hire. But that doesn't quite feedback to the ultimate goal of having oodles of cash as academic pay has a natural limit.

And the main task though is to actually be expert in this area, rather than just appearing as one! Feb will see the DataCamp specialism completed, and a fair bit of work on datasets to build up core pandas knowledge (the building blocks of data science). Then March onwards is all about building up proper specialism in machine learning: the ultimate goal. The book chapter and article will be great in this regard as chance to understand where the area is and think about where it's going.

Last edited by Hitchhiker's Guide To...; 30-01-18 at 23:51.
Hitchhiker's Guide To... is offline   Reply With Quote
3 Thanks From:
Old 30-01-18, 23:48   #31
Mellor
Member
 
Mellor's Avatar
 
Join Date: Jan 2010
Location: Location: Location:
Posts: 10,316
I've nothing to add except I like that title
I'm not even sure what python is for
__________________
Mellor is offline   Reply With Quote
Old 30-01-18, 23:54   #32
Hitchhiker's Guide To...
Duncan F*****g Stewart
 
Hitchhiker's Guide To...'s Avatar
 
Join Date: Jan 2010
Location: South Cidre
Posts: 28,764
The new Wes McKinney 'Python for Data Analysis' is out now for all the pandas insights needed. Seems like a very good book. Going to work through it in February as well, as a lot will be known.
Hitchhiker's Guide To... is offline   Reply With Quote
Old 31-01-18, 06:08   #33
brady23
Member
 
Join Date: Apr 2010
Posts: 2,210
I started the Applied Data Science with Python from Michigan this week and I hope to have the VBA done by next week.
It's been suggested to make life easier to use them for 2 of the classes but you could struggle along with excel for both.
The only disappointing part is that neither will be examinable but it'll simply make the project work easier.
Semester 2 feels incredibly short, last week of classes finishes the 23rd of March
brady23 is offline   Reply With Quote
Old 31-01-18, 08:33   #34
Hitchhiker's Guide To...
Duncan F*****g Stewart
 
Hitchhiker's Guide To...'s Avatar
 
Join Date: Jan 2010
Location: South Cidre
Posts: 28,764
Quote:
Originally Posted by brady23 View Post
I started the Applied Data Science with Python from Michigan this week and I hope to have the VBA done by next week.
It's been suggested to make life easier to use them for 2 of the classes but you could struggle along with excel for both.
The only disappointing part is that neither will be examinable but it'll simply make the project work easier.
Semester 2 feels incredibly short, last week of classes finishes the 23rd of March
That's mad quick. What are you doing the dissertation on?
Hitchhiker's Guide To... is offline   Reply With Quote
Old 31-01-18, 08:51   #35
brady23
Member
 
Join Date: Apr 2010
Posts: 2,210
Quote:
Originally Posted by Hitchhiker's Guide To... View Post
That's mad quick. What are you doing the dissertation on?
I'm hoping something in the area of prospect theory and physiological response.
They seem to think it's a good idea although it's a bit off topic so I'll know once the first draft of the research proposal is submitted.

I figure I'm not going to make any contribution in the area of finance by determining historical correlation of x stock etc plus it's not something that particularly interests me.
I don't expect to make any particular contribution either way but I do think this topic has far more scope for (dare I say it) a PhD.

Everybody is going to try cryptos I'm sure or work with Brian on his IPO stuff (which probably makes the most sense tbf).

I'll send you a copy of the research proposal when it's completed and get your thoughts, I know behavioural is your area.
brady23 is offline   Reply With Quote
Old 31-01-18, 10:17   #36
DeadParrot
Grand High Wizard
 
DeadParrot's Avatar
 
Join Date: Jan 2010
Posts: 18,643
Hitch,

just a quick thing. What I have noticed in here is the python guys struggle with static offline data sources as python does the heavy lifting opening csvs etc.
Use a proper DB engine as a data source and performance is markedly improved.
I'm still level 0.1 in python but hoping to improve
__________________
People say I should be more humble
I hope they understand, they don't listen when you mumble
The OnePlus 6, when you absolutely have to have the best phone in the room, accept no substitutes.
get 20 towards accessories or discount one your phone on @ oneplus.net. We all win
DeadParrot is offline   Reply With Quote
Old 31-01-18, 19:00   #37
Hitchhiker's Guide To...
Duncan F*****g Stewart
 
Hitchhiker's Guide To...'s Avatar
 
Join Date: Jan 2010
Location: South Cidre
Posts: 28,764
Quote:
Originally Posted by brady23 View Post
I'm hoping something in the area of prospect theory and physiological response.
They seem to think it's a good idea although it's a bit off topic so I'll know once the first draft of the research proposal is submitted.

I figure I'm not going to make any contribution in the area of finance by determining historical correlation of x stock etc plus it's not something that particularly interests me.
I don't expect to make any particular contribution either way but I do think this topic has far more scope for (dare I say it) a PhD.

Everybody is going to try cryptos I'm sure or work with Brian on his IPO stuff (which probably makes the most sense tbf).

I'll send you a copy of the research proposal when it's completed and get your thoughts, I know behavioural is your area.
Yeah do that. I'm working on a prospect theory paper at the moment, nearly finished. I presume its the Indian guy who will be your supervisor so. Read the prospect theory paper he did with Brian.


Quote:
Originally Posted by DeadParrot View Post
Hitch,

just a quick thing. What I have noticed in here is the python guys struggle with static offline data sources as python does the heavy lifting opening csvs etc.
Use a proper DB engine as a data source and performance is markedly improved.
I'm still level 0.1 in python but hoping to improve
That very much matches with some things I've read, but then keep forgetting! Are you talking sql or something else like that, or something else altogether?
Hitchhiker's Guide To... is offline   Reply With Quote
Old 01-02-18, 11:02   #38
DeadParrot
Grand High Wizard
 
DeadParrot's Avatar
 
Join Date: Jan 2010
Posts: 18,643
Any proper DB engine MSSQL for me anyway
__________________
People say I should be more humble
I hope they understand, they don't listen when you mumble
The OnePlus 6, when you absolutely have to have the best phone in the room, accept no substitutes.
get 20 towards accessories or discount one your phone on @ oneplus.net. We all win
DeadParrot is offline   Reply With Quote
Old 01-02-18, 14:15   #39
Hitchhiker's Guide To...
Duncan F*****g Stewart
 
Hitchhiker's Guide To...'s Avatar
 
Join Date: Jan 2010
Location: South Cidre
Posts: 28,764
Cheers. Must look into that. Today's course is some basics around that idea - using sqlalchemy in python to read, extract, write to a range of sql dbs. But mainly sqlite.
Hitchhiker's Guide To... is offline   Reply With Quote
Old 01-02-18, 14:51   #40
Elshambles
Wierdo
 
Elshambles's Avatar
 
Join Date: Jan 2010
Location: Turning on ALL the taps
Posts: 7,606
Thinking of making python part of my time doing courses

What would you advise to someone coming at it from a fado fado coding viewpoint not a financial one?

Ta
__________________
PMA
Elshambles is offline   Reply With Quote
Reply

  Irish Poker Boards > Community, Tech, and Feedback > Personal Logs

Bookmarks

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off



All times are GMT. The time now is 02:39.