I want 10 hours of work done coding in R.
I have a set of clear instructions and existing code (5 modules, ~1000 lines including comments).
I need the code commented and altered based on the instructions.
The alterations include:
- reverse engineering parameters for function calls
- removing link to database and replacing with data frames and tables in R
VINEGAR TASKS FOR THE FIRST SPRINT
I am passing you 4 .R files, an example input data file, as well as this text file.
TASK 1: UNDERSTAND THE CODE
The first task is to scan the main R file, vinegar.R.
The original ancient comments are in there, but I've also added a bunch of new comments
prefixed with "#M@171122: "
Read all of my comments, then have a better look at the code.
You will see the parts that refer to the database so have a quick look at that code in
DB-OPERATIONS.R and my comments.
Basically the database it used to work with is so ugly!! It had way too many tables, a whole
redundant concept of different "batches", where each batch had different time series and the
ability to prioritise different runs, different features, different functions. Please don't spend too
much time trying to figure this stuff out, just rewrite the data in and data out models and ignore
this function. I've only included it for you to perhaps look at how parameters are set out I
wouldn't trust it however, possibly this function is misaligned with the main code you are working
Then have a look through modelling.R and any other remaining code.
TASK 2 : REVERSE THE DATA MODEL AND "HARD CALL" WITH R A SINGLE RUN THAT
There were parameters for the different:
● time series
● their "IDs" (which is redundant)
● smoothing options
● the modelling options (multiple per run), and the lagging
● parameters for the different tolerances to continue (i.e. how well the models had to
perform at various "abandon" stages
(see line 113 of vinegar.r, however the content of param.str might not be 100% correct, it’s just
Task 2.1 As an input to the code, you should set up a data frame with 3 tables:
1. Time series data #1 (two columns - time date stamp and a value)
2. Time series data #2 (two columns - time date stamp and a value)
3. A parameters list for the run
The parameters list / permutations needs to be derived from the code - i.e. you
need to reverse engineering this. This includes smoothing, modelling technique,
thresholds to proceed, the minimum and maximum lags, etc.
Task 2.2 See if you can get a single simple model to run, e.g. no smoothing, modelling
LR only with no lags.
Task 2.3 Then try out the features incrementally (smoothing, lagging etc.). This will be
done by changing the parameters in table 3 above.
Task 2.4 Please add comments and document what the data structures are as you figure
Task 2.5 As various points in vinegar.R, e.g. line 330, it writes the results back into the
connected database. Please remove usage of the database, and instead as an
output to the code, you should set up whatever data frames and tables you
need, and write to them.
Task 2.6 The code also outputs plots, etc. as JPG, we want these to be connected to the
modelling results, and we want to understand what is being generated there.
14 freelancers estão ofertando em média $541 para esse trabalho
Hello I can achieve ths project perfectly using R I master r coding data mining statistics analysis machine learning please contact me for more details about the project best regards
Hi, I am a data scientist who uses R a lot. I think we can discuss more about the detail of the project. Look forward to working with you. All the best, Worrawat