How can I use the R packages zoo or xts with very large data sets? (100GB)
I know there are some packages such as bigrf, ff, bigmemory that can deal with this problem but you have to use their limited set of commands, they don't have the functions of zoo or xts and I don't know how to make zoo or xts to use them.
How can I use it?
I've seen that there are also some other things, related with databases, such as sqldf and hadoopstreaming, RHadoop, or some other used by Revolution R. What do you advise?, any other?
I just want to aggreagate series, cleanse, and perform some cointegrations and plots.
I wouldn't like to need to code and implement new functions for every command I need, using small pieces of data every time.
Added: I'm on Windows
I have had a similar problem (albeit I was only playing with 9-10 GBs). My experience is that there is no way R can handle so much data on its own, especially since your dataset appears to contain time series data.
If your dataset contains a lot of zeros, you may be able to handle it using sparse matrices - see Matrix package ( http://cran.r-project.org/web/packages/Matrix/index.html ); this manual may also come handy ( http://www.johnmyleswhite.com/notebook/2011/10/31/using-sparse-matrices-in-r/ )
I used PostgreSQL - the relevant R package is RPostgreSQL ( http://cran.r-project.org/web/packages/RPostgreSQL/index.html ). It allows you to query your PostgreSQL database; it uses SQL syntax. Data is downloaded into R as a dataframe. It may be slow (depending on the complexity of your query), but it is robust and can be handy for data aggregation.
Drawback: you would need to upload data into the database first. Your raw data needs to be clean and saved in some readable format (txt/csv). This is likely to be the biggest issue if your data is not already in a sensible format. Yet uploading "well-behaved" data into the DB is easy ( see http://www.postgresql.org/docs/8.2/static/sql-copy.html and How to import CSV file data into a PostgreSQL table? )
I would recommend using PostgreSQL or any other relational database for your task. I did not try Hadoop, but using CouchDB nearly drove me round the bend. Stick with good old SQL
Related
While working with R I encountered a strange problem:
I am processing date in the follwing manner:
Reading data from a database into a dataframe, filling missing values, grouping and nesting the data to a combined primary key, creating a timeseries and forecastting it for every group, ungroup and clean the data, write it back into the DB.
Somehting like this:
https://cran.rstudio.com/web/packages/sweep/vignettes/SW01_Forecasting_Time_Series_Groups.html
For small data sets this works like a charm, but with lager ones (over about 100000 entries) I do get the "R Session Aborted" screen from R-Studio and the nativ R GUI just stops execution and implodes.
There is no information in every log file that I've look into. I suspect that it is some kind of (leaking) memory issue.
As a work around I'm processing the data in chunks with a for-loop. But no matter how small the chunk size is, I do get the "R Session Aborted" screen, which looks a lot like leaking memory.
The whole date consist of about 5 million rows.
I've looked a lot into packages like ff, the big-Family and matter basically everything from https://cran.r-project.org/web/views/HighPerformanceComputing.html
but this dose not seem to work well with tibbles and the tidyverse way of data processing.
So, how can I improve my scrip to work with massiv amounts of data?
How can I gather clues about why the R Session is Aborted?
Check out the article at:
datascience.la/dplyr-and-a-very-basic-benchmark
There is a table that shows runtime comparisons for some of the data wrangling tasks you are performing. From the table, it looks as though dplyr with data.table behind it is likely going to do much better than dplyr with a dataframe behind it.
There’s a link to the benchmarking code used to make the table, too.
In short, try adding a key, and try using data.table over dataframe.
To make x your key, and say your data.table is named dt, use setkey(dt,x).
While Pakes answer deals with the described problem I found a solution to the underlying problem. For Compatibility reason I used R in the 3.4.3 version. Now I'm using the newer 3.5.1 version which works quite fine.
I am using the package RMySQL with DBI package in R.
When I run the code,
dbReadTable(con, "data")
it is taking forever.
I think the table is very big data.
Any ideas on how to speed up this process?
Thanks,
Try to get the database to do as much filtering & processing as possible. A database has many more ways to optimize operations than R, and isn't constrained by RAM so severely. It also reduces the amount that has to travel across the network.
Common approaches tactics are
using the WHERE clause to reduce rows
explicitly list (only the necessary) columns, instead of using *
do as much aggregation in SQL as possible (eg, GROUP BY + MAX)
use INSERT queries to write from table to table, so the data doesn't even need to pass through R.
I imagine RMySQL should be faster than the newish odbc package, but it's worth experimenting with.
What's 'forever'? 5 min or 5 hours? Are things still slow once the data get to R? If things are still too slow to be feasible, consider escalating to something like sparklyr.
I have a 500K row spark DataFrame that lives in a parquet file. I'm using spark 2.0.0 and the SparkR package inside Spark (RStudio and R 3.3.1), all running on a local machine with 4 cores and 8gb of RAM.
To facilitate construction of a dataset I can work on in R, I use the collect() method to bring the spark DataFrame into R. Doing so takes about 3 minutes, which is far longer than it'd take to read an equivalently sized CSV file using the data.table package.
Admittedly, the parquet file is compressed and the time needed for decompression could be part of the issue, but I've found other comments on the internet about the collect method being particularly slow, and little in the way of explanation.
I've tried the same operation in sparklyr, and it's much faster. Unfortunately, sparklyr doesn't have the ability to do date path inside joins and filters as easily as SparkR, and so I'm stuck using SparkR. In addition, I don't believe I can use both packages at the same time (i.e. run queries using SparkR calls, and then access those spark objects using sparklyr).
Does anyone have a similar experience, an explanation for the relative slowness of SparkR's collect() method, and/or any solutions?
#Will
I don't know whether the following comment actually answers your question or not but Spark does lazy operations. All the transformations done in Spark (or SparkR) doesn't really create any data they just create a logical plan to follow.
When you run Actions like collect, it has to fetch data directly from source RDDs (assuming you haven't cached or persisted data).
If your data is not large enough and can be handled by local R easily then there is no need for going with SparkR. Other solution can be to cache your data for frequent uses.
Short: Serialization/deserialization is very slow.
See for example post on my blog http://dsnotes.com/articles/r-read-hdfs
However it should be equally slow in both sparkR and sparklyr.
What is the best format to persist simple data frames to disc in R for storage while limiting semantic loss?
I ask because I'm archiving a data set. In an ideal world, my data format would have the follow characteristics:
Stability - the storage format will be compatible with future version of R
Semantic compatibility - the storage format will understand the semantics of R's primative data types. For example, it will be able to store ordered factors with labels in a sensible manner.
Open standard - ideally, the format will be an open standard so other statistics packages (now or in the future) will be able to understand it
My first thought was to use CSV which is very stable, but lacks the semantic richness required. On the other hand, R's builtin RData format completely captures R's semantics, but seems likely to change between releases (correct me if I'm wrong).
Is there another format that finds a balance between these three imperatives?
Dump it to a text file with dput. That way you get all the structure of R's objects, and its in a text-based form that, should R stop existing, can be parsed fairly easily.
It probably doesn't pass (3), your 'open standard' test.
R is pretty good for backward compatibility with its .RData format, so even if the files written by the latest R aren't the same as older ones, the latest R will still read old files. However, if R should stop existing, reverse-engineering of the binary format is orders of magnitude harder than grokking the output from dput.
I know this is not a new concept by any stretch in R, and I have browsed the High Performance and Parallel Computing Task View. With that said, I am asking this question from a point of ignorance as I have no formal training in Computer Science and am entirely self taught.
Recently I collected data from the Twitter Streaming API and currently the raw JSON sits in a 10 GB text file. I know there have been great strides in adapting R to handle big data, so how would you go about this problem? Here are just a handful of the tasks that I am looking to do:
Read and process the data into a data frame
Basic descriptive analysis, including text mining (frequent terms, etc.)
Plotting
Is it possible to use R entirely for this, or will I have to write some Python to parse the data and throw it into a database in order to take random samples small enough to fit into R.
Simply, any tips or pointers that you can provide will be greatly appreciated. Again, I won't take offense if you describe solutions at a 3rd grade level either.
Thanks in advance.
If you need to operate on the entire 10GB file at once, then I second #Chase's point about getting a larger, possibly cloud-based computer.
(The Twitter streaming API returns a pretty rich object: a single 140-character tweet could weigh a couple kb of data. You might reduce memory overhead if you preprocess the data outside of R to extract only the content you need, such as author name and tweet text.)
On the other hand, if your analysis is amenable to segmenting the data -- for example, you want to first group the tweets by author, date/time, etc -- you could consider using Hadoop to drive R.
Granted, Hadoop will incur some overhead (both cluster setup and learning about the underlying MapReduce model); but if you plan to do a lot of big-data work, you probably want Hadoop in your toolbox anyway.
A couple of pointers:
an example in chapter 7 of Parallel R shows how to setup R and Hadoop for large-scale tweet analysis. The example uses the RHIPE package, but the concepts apply to any Hadoop/MapReduce work.
you can also get a Hadoop cluster via AWS/EC2. Check out
Elastic MapReduce
for an on-demand cluster, or use
Whirr
if you need more control over your Hadoop deployment.
There's a brand-new package called colbycol that lets you read in only the variables you want from enormous text files:
http://colbycol.r-forge.r-project.org/
read.table function remains the main data import function in R. This
function is memory inefficient and, according to some estimates, it
requires three times as much memory as the size of a dataset in order
to read it into R.
The reason for such inefficiency is that R stores data.frames in
memory as columns (a data.frame is no more than a list of equal length
vectors) whereas text files consist of rows of records. Therefore, R's
read.table needs to read whole lines, process them individually
breaking into tokens and transposing these tokens into column oriented
data structures.
ColByCol approach is memory efficient. Using Java code, tt reads the
input text file and outputs it into several text files, each holding
an individual column of the original dataset. Then, these files are
read individually into R thus avoiding R's memory bottleneck.
The approach works best for big files divided into many columns,
specially when these columns can be transformed into memory efficient
types and data structures: R representation of numbers (in some
cases), and character vectors with repeated levels via factors occupy
much less space than their character representation.
Package ColByCol has been successfully used to read multi-GB datasets
on a 2GB laptop.
10GB of JSON is rather inefficient for storage and analytical purposes. You can use RJSONIO to read it in efficiently. Then, I'd create a memory mapped file. You can use bigmemory (my favorite) to create different types of matrices (character, numeric, etc.), or store everything in one location, e.g. using HDF5 or SQL-esque versions (e.g. see RSQlite).
What will be more interesting is the number of rows of data and the number of columns.
As for other infrastructure, e.g. EC2, that's useful, but preparing a 10GB memory mapped file doesn't really require much infrastructure. I suspect you're working with just a few 10s of millions of rows and a few columns (beyond the actual text of the Tweet). This is easily handled on a laptop with efficient use of memory mapped files. Doing complex statistics will require either more hardware, cleverer use of familiar packages, and/or experimenting with some unfamiliar packages. I'd recommend following up with a more specific question when you reach that stage. The first stage of such work is simply data normalization, storage and retrieval. My answer for that is simple: memory mapped files.
To read chunks of the JSON file in, you can use the scan() function. Take a look at the skip and nlines arguments. I'm not sure how much performance you'll get versus using a database.