How are you assigned tasks? Can you give an example you struggled with? Are they more orientated around getting the job done than production level code?

It's pretty hard to get rid of someone at least over here.

I definitely want the job! I really like it, just need to keep grinding and improving.

About 9 months. And yeah it's normal to look things up. I just feel like I shouldn't have to do it constantly.

He mentioned how he could do a task that was taking me a while in 30 minutes and I should be able to as well.

Well, an idea is devised by someone on how we can increase revenue or cut costs using data science. And the project takes off from there. It's not structured unless we are taking part in some sort of program that stretches beyond our team. In that case, our goals and projects will be predefined.

An example of a struggle would be wanting to wanting to scrape JSON data from twitters API and doing some sort of textual analysis. I'm fine at conceptualizing what needs to be done. I'm fine at understanding what algorithms to use in certain cases. But I may well struggle at writing the code to parse the JSON data. And then once I finally do that I may struggle with writing the code to, say, cluster the tweets based on sentiments expressed in the tweets.

As for the last question, it's about getting the job done. We don't "make things." We don't send anything to production. We obtain data, transform it, analyze it, and share the output with whomever it is that would benefit from knowing the results of the analysis.

I know of a few o'reilly books that cover exactly that type of task. One is called 'mining the social web' and another is called 'collective intelligence'. They both use python. You should learn python. If you're struggling you really want to be able to find tutorials/stackoverflows to help you along & python probably has more resources. Also it doesn't sound like you're doing anything too academic so there are probably scipi type libraries with algorithm implementations already written. You're probably wasting a ton of time reinventing the wheel to write what amounts to 10 lines of python.

Yeah I'm using R now but really want to learn Python. Those books sound intriguing, I'll have to look into those. Thanks.

Honest question - Over the last week what have you done to address your weaknesses?

Good question JJ. My main source of development right now is coursera's data science specialization track. It consists of 9 four week courses and a capstone project. I'm about to wrap up the third course. These courses are taught in R.

I also meet with a more advanced colleague once a week but this is more of an "address a current problem I'm having" meeting than a structured learning thing.

I feel like I should do more. Perhaps reading and coding in Python twice a week in my spare time or something like that.

Do you get home really late or have no time at the end of the day by being busy or tired?

I just don't see how coding twice a week is making much progress in learning or attempting much. How long are the two sessions?

Also what are you building in the learning sessions.

Well that would be in addition to the coursera coursework I do on a weekly basis. I workout three times a week, Brazilian jiu-jitsu twice a week, have a gf and active social life. Maybe I just need to sacrifice some of my health/social life for the near future?

In the coursera sessions, I'm doing various data analysis tasks; I'm not really building anything. I write code that does some combination of acquire/transform/analyze data. Data can consist of anything from an excel file to a webpage's HTML.

If I coded in python on the side, again, I'm not sure I'd build anything. I'd try to figure out the best way to learn python for data analysis and if that involved building something, then so be it.

The reason I posted about thinking hard about where you're weak is that you want to use your limited time to address your most pressing issues.

This doesn't seem like its helped too much by taking Coursera sessions. It sounds like that would be helped by just building stuff and solving example problems.

I'm not suggesting you ruin your social life because that is something more valuable than work.

Just maybe consider that the way you're trying to learn is not optimal and that looking at code every day will help your brain pick it up faster. Same with trying to learn new things you can do with code.

Ok cool. Good advice guys. Thanks a lot. Would something like codeacademy be good for getting me coding more? Any recommendations? Should I try to specifically learn it in the context of my job or just learn the language in general?

I will suggest ditching your social life. As Mr Burns says: Friends, family, religion - these are the three demons you must slay if you want to be successful.

Not really being that serious, but if you can go easy on going out and spend more time learning more skills for work , the payoff in the end will be very worth it.

You could also go insane but what are the chances of that.

It depends. If you really don't feel you're having problems with the data science part I think there's a lot of value in just picking various data sets and trying do something interesting with them. Ideally, the same sort of things that you need to do in real life (perhaps even your real data).

For example, I recently used scikit-learn's newsgroup dataset (http://scikit-learn.org/stable/datas...ewsgroups.html) to play around with some online learning stuff. In this case I just wanted to learn some data science stuff and so a small dataset that I could iterate over quickly was ideal.

If you want to be efficient with your time, you need to have a really good idea of where you're going wrong and figure out how to work on that specifically without wasting a bunch of time on other stuff. As you get better, move on to other things.

I think regular code reviews - ie. you reviewing your peers' and more experienced co-workers' code - would be very beneficial.
And not in the sense of "I shall admire whatever you produce" but rather seriously trying to find spots where his code could improve. That will lead to a set of questions that may teach you more about relevant coding skills in less time.
Vice versa will obviously help, too.

Also, a lot of the tasks should be repetitive, so building and sharing a small library for general tasks such as scraping data should be beneficial. These can evolve over time; just don't try to expand functionality beyond the core tasks these packages are designed for - there be dragons, and no pretty ones.

I appreciate all of the input guys. The theme of the thread seems to be all about the code. I need to code more. Having a skilled colleague review my code is easy to achieve. I'm heading on a quick 4 day vacation tomorrow but will be looking for resources soon and perhaps I'll ask for ya'lls thoughts on them.

I really agree with this; being put into a position where you and someone else are looking at same code is always beneficial.

Social sharing of work speeds learning and sharing techniques that are valuable.

The worst thing you can do as a company is hire someone new and not have him working with others, which is sort of like taking a gamble on if the person will deliver.

pfft, i only hang out with people like twice a year

it's very overrated

Yeah I agree with this. All of our projects are more or less individual. I still have colleagues look at my code sometimes but they aren't working in it with me so it's not the same.

Are you looking at their code? In a critical way?

I'm not. If I asked for some of their code, how would you recommend critiquing it?

Simple be blunt would be fastest way to find out.

Tell them you want to learn by looking at anything that someone will give to you, any sort of project that would be useful for your work and you could utilize.

You will most likely think of some changes when you understand the code or maybe you won't but you will run into something in the future.