Related
First, sorry if the title is not clear. I (noob) am baffled by this...
Here's my code:
import csv
from random import random
from collections import Counter
def rn(dic, p):
for ptry in parties:
if p < float(dic[ptry]):
return ptry
else:
p -= float(dic[ptry])
def scotland(r):
r['SNP'] = 48
r['Con'] += 5
r['Lab'] += 1
r['LibDem'] += 5
def n_ireland(r):
r['DUP'] = 9
r['Alliance'] = 1
# SF = 7
def election():
results = Counter([rn(row, random()) for row in data])
scotland(results)
n_ireland(results)
return results
parties = ['Con', 'Lab', 'LibDem', 'Green', 'BXP', 'Plaid', 'Other']
with open('/Users/andrew/Downloads/msp.csv', newline='') as f:
data = csv.DictReader(f)
for i in range(1000):
print(election())
What happens is that in every iteration after the first one, the variable data seems to have vanished: the function election() creates a Counter object from a list obtained by processing data, but on every pass after the first one, this object is empty, so the function just returns the hard coded data from scotland() and n_ireland(). (msp.csv is a csv file containing detailed polling data). I'm sure I'm doing something stupid but would welcome anyone gently pointing out where...
I’m going to place a bet on your definition of newline. Are you sure you don’t want newline = “\n” ? Otherwise it will interpret the entire file as a single line, which explains what you’re seeing.
EDIT
I now see another issue. The file object in python acts as a generator for each line. The problem is once the generator is finished (you hit the end of the file), you have no more data generated. To solve this: reset your file pointer to the beginning of the file like so:
with open('/Users/andrew/Downloads/msp.csv') as f:
data = csv.DictReader(f)
for i in range(1000):
print(election())
f.seek(0)
Here the call to f.seek(0) will reset the file pointer to the beginning of your file. You are correct that data is a global object given the way you've defined it at the module level, there's no need to pass it as a parameter.
I agree with #smassey, you might need to change the code to
with open('/Users/andrew/Downloads/msp.csv', newline='\n') as f:
or simply try not use that argument
with open('/Users/andrew/Downloads/msp.csv') as f:
I have a large CSV file full of stock-related data formatted as such:
Ticker Symbol, Date, [some variables...]
So each line starts of with the symbol (like "AMZN"), then has the date, then has 12 variables related to price or volume on the selected date. There are about 10,000 different securities represented in this file and I have a line for each day that the stock has been publicly traded for each of them. The file is ordered first alphabetically by ticker symbol and second chronologically by date. The entire file is about 3.3 GB.
The sort of task I want to solve would be to be able to extract the most recent n lines of data for a given ticker symbol with respect to the current date. I have code that does this, but based on my observations it seems to take, on average, around 8-10 seconds per retrieval (all tests have been extracting 100 lines).
I have functions I'd like to run that require me to grab such chunks for hundreds or thousands of symbols, and I would really like to reduce the time. My code is inefficient, but I am not sure how to make it run faster.
First, I have a function called getData:
def getData(symbol, filename):
out = ["Symbol","Date","Open","High","Low","Close","Volume","Dividend",
"Split","Adj_Open","Adj_High","Adj_Low","Adj_Close","Adj_Volume"]
l = len(symbol)
beforeMatch = True
with open(filename, 'r') as f:
for line in f:
match = checkMatch(symbol, l, line)
if beforeMatch and match:
beforeMatch = False
out.append(formatLineData(line[:-1].split(",")))
elif not beforeMatch and match:
out.append(formatLineData(line[:-1].split(",")))
elif not beforeMatch and not match:
break
return out
(This code has a couple of helper functions, checkMatch and formatLineData, which I will show below.) Then, there is another function called getDataColumn that gets the column I want with the correct number of days represented:
def getDataColumn(symbol, col=12, numDays=100, changeRateTransform=False):
dataset = getData(symbol)
if not changeRateTransform:
column = [day[col] for day in dataset[-numDays:]]
else:
n = len(dataset)
column = [(dataset[i][col] - dataset[i-1][col])/dataset[i-1][col] for i in range(n - numDays, n)]
return column
(changeRateTransform converts raw numbers into daily change rate numbers if True.) The helper functions:
def checkMatch(symbol, symbolLength, line):
out = False
if line[:symbolLength+1] == symbol + ",":
out = True
return out
def formatLineData(lineData):
out = [lineData[0]]
out.append(datetime.strptime(lineData[1], '%Y-%m-%d').date())
out += [float(d) for d in lineData[2:6]]
out += [int(float(d)) for d in lineData[6:9]]
out += [float(d) for d in lineData[9:13]]
out.append(int(float(lineData[13])))
return out
Does anyone have any insight on what parts of my code run slow and how I can make this perform better? I can't do the sort of analysis I want to do without speeding this up.
EDIT:
In response to the comments, I made some changes to the code in order to utilize the existing methods in the csv module:
def getData(symbol, database):
out = ["Symbol","Date","Open","High","Low","Close","Volume","Dividend",
"Split","Adj_Open","Adj_High","Adj_Low","Adj_Close","Adj_Volume"]
l = len(symbol)
beforeMatch = True
with open(database, 'r') as f:
databaseReader = csv.reader(f, delimiter=",")
for row in databaseReader:
match = (row[0] == symbol)
if beforeMatch and match:
beforeMatch = False
out.append(formatLineData(row))
elif not beforeMatch and match:
out.append(formatLineData(row))
elif not beforeMatch and not match:
break
return out
def getDataColumn(dataset, col=12, numDays=100, changeRateTransform=False):
if not changeRateTransform:
out = [day[col] for day in dataset[-numDays:]]
else:
n = len(dataset)
out = [(dataset[i][col] - dataset[i-1][col])/dataset[i-1][col] for i in range(n - numDays, n)]
return out
Performance was worse using the csv.reader class. I tested on two stocks, AMZN (near top of file) and ZNGA (near bottom of file). With the original method, the run times were 0.99 seconds and 18.37 seconds, respectively. With the new method leveraging the csv module, the run times were 3.04 seconds and 64.94 seconds, respectively. Both return the correct results.
My thought is that the time is being taken up more from finding the stock than from the parsing. If I try these methods on the first stock in the file, A, the methods both run in about 0.12 seconds.
When you're going to do lots of analysis on the same dataset, the pragmatic approach would be to read it all into a database. It is made for fast querying; CSV isn't. Use the sqlite command line tools, for example, which can directly import from CSV. Then add a single index on (Symbol, Date) and lookups will be practically instantaneous.
If for some reason that is not feasible, for example because new files can come in at any moment and you cannot afford the preparation time before starting your analysis of them, you'll have to make the best of dealing with CSV directly, which is what the rest of my answer will focus on. Remember that it's a balancing act, though. Either you pay a lot upfront, or a bit extra for every lookup. Eventually, for some amount of lookups it would have been cheaper to pay upfront.
Optimization is about maximizing the amount of work not done. Using generators and the built-in csv module aren't going to help much with that in this case. You'd still be reading the whole file and parsing all of it, at least for line breaks. With that amount of data, it's a no-go.
Parsing requires reading, so you'll have to find a way around it first. Best practices of leaving all intricacies of the CSV format to the specialized module bear no meaning when they can't give you the performance you want. Some cheating must be done, but as little as possible. In this case, I suppose it is safe to assume that the start of a new line can be identified as b'\n"AMZN",' (sticking with your example). Yes, binary here, because remember: no parsing yet. You could scan the file as binary from the beginning until you find the first line. From there read the amount of lines you need, decode and parse them the proper way, etc. No need for optimization there, because a 100 lines are nothing to worry about compared to the hundreds of thousands of irrelevant lines you're not doing that work for.
Dropping all that parsing buys you a lot, but the reading needs to be optimized as well. Don't load the whole file into memory first and skip as many layers of Python as you can. Using mmap lets the OS decide what to load into memory transparently and lets you work with the data directly.
Still you're potentially reading the whole file, if the symbol is near the end. It's a linear search, which means the time it takes is linearly proportional to the number of lines in the file. You can do better though. Because the file is sorted, you could improve the function to instead perform a kind of binary search. The number of steps that will take (where a step is reading a line) is close to the binary logarithm of the number of lines. In other words: the number of times you can divide your file into two (almost) equally sized parts. When there are one million lines, that's a difference of five orders of magnitude!
Here's what I came up with, based on Python's own bisect_left with some measures to account for the fact that your "values" span more than one index:
import csv
from itertools import islice
import mmap
def iter_symbol_lines(f, symbol):
# How to recognize the start of a line of interest
ident = b'"' + symbol.encode() + b'",'
# The memory-mapped file
mm = mmap.mmap(f.fileno(), 0, access=mmap.ACCESS_READ)
# Skip the header
mm.readline()
# The inclusive lower bound of the byte range we're still interested in
lo = mm.tell()
# The exclusive upper bound of the byte range we're still interested in
hi = mm.size()
# As long as the range isn't empty
while lo < hi:
# Find the position of the beginning of a line near the middle of the range
mid = mm.rfind(b'\n', 0, (lo+hi)//2) + 1
# Go to that position
mm.seek(mid)
# Is it a line that comes before lines we're interested in?
if mm.readline() < ident:
# If so, ignore everything up to right after this line
lo = mm.tell()
else:
# Otherwise, ignore everything from right before this line
hi = mid
# We found where the first line of interest would be expected; go there
mm.seek(lo)
while True:
line = mm.readline()
if not line.startswith(ident):
break
yield line.decode()
with open(filename) as f:
r = csv.reader(islice(iter_symbol_lines(f, 'AMZN'), 10))
for line in r:
print(line)
No guarantees about this code; I didn't pay much attention to edge cases, and I couldn't test with (any of) your file(s), so consider it a proof of concept. It is plenty fast, however – think tens of milliseconds on an SSD!
So I have an alternative solution which I ran and tested on my own as well with a sample data set that I got on Quandl that appears to have all the same headers and similar data. (Assuming that I havent misunderstood the end result that your trying to achieve).
I have this command line tool that one of our engineers built for us for parsing massive csvs - since I deal with absurd amount of data on a day to day basis - it is open sourced and you can get it here: https://github.com/DataFoxCo/gocsv
I also already wrote the short bash script for it in case you don't want to pipeline the commands but it does also support pipelining.
The command to run the following short script follows a super simple convention:
bash tickers.sh wikiprices.csv 'AMZN' '2016-12-\d+|2016-11-\d+'
#!/bin/bash
dates="$3"
cat "$1" \
| gocsv filter --columns 'ticker' --regex "$2" \
| gocsv filter --columns 'date' --regex "$dates" > "$2"'-out.csv'
both arguments for ticker and for dates are regexes
You can add as many variations as your want into that one regex, separating them by |.
So if you wanted AMZN and MSFT then you would simply modify it to this: AMZN|MSFT
I did something very similar with the dates - but i only limited my sample run to any dates from this month or last month.
End Result
Starting data:
myusername$ gocsv dims wikiprices.csv
Dimensions:
Rows: 23946
Columns: 14
myusername$ bash tickers.sh wikiprices.csv 'AMZN|MSFT' '2016-12-\d+'
myusername$ gocsv dims AMZN|MSFT-out.csv
Dimensions:
Rows: 24
Columns: 14
Here is a sample where I limited to only those 2 tickers and then to december only:
Voila - in a matter of seconds you have a second file saved with out the data you care about.
The gocsv program has great documentation by the way - and a ton of other functions e.g. running a vlookup basically at any scale (which is what inspired the creator to make the tool)
in addition to using csv.reader I think using itertools.groupby would speed up looking for the wanted sections, so the actual iteration could look something like this:
import csv
from itertools import groupby
from operator import itemgetter #for the keyfunc for groupby
def getData(wanted_symbol, filename):
with open(filename) as file:
reader = csv.reader(file)
#so each line in reader is basically line[:-1].split(",") from the plain file
for symb, lines in groupby(reader, itemgetter(0)):
#so here symb is the symbol at the start of each line of lines
#and lines is the lines that all have that symbol in common
if symb != wanted_symbol:
continue #skip this whole section if it has a different symbol
for line in lines:
#here we have each line as a list of fields
#for only the lines that have `wanted_symbol` as the first element
<DO STUFF HERE>
so in the space of <DO STUFF HERE> you could have the out.append(formatLineData(line)) to do what your current code does but the code for that function has a lot of unnecessary slicing and += operators which I think are pretty expensive for lists (might be wrong), another way you could apply the conversions is to have a list of all the conversions:
def conv_date(date_str):
return datetime.strptime(date_str, '%Y-%m-%d').date()
#the conversions applied to each element (taken from original formatLineData)
castings = [str, conv_date, #0, 1
float, float, float, float, #2:6
int, int, int, #6:9
float, float, float, float, #9:13
int] #13
then use zip to apply these to each field in a line in a list comprehension:
[conv(val) for conv, val in zip(castings, line)]
so you would replace <DO STUFF HERE> with out.append with that comprehension.
I'd also wonder if switching the order of groupby and reader would be better since you don't need to parse most of the file as csv, just the parts you are actually iterating over so you could use a keyfunc that seperates just the first field of the string
def getData(wanted_symbol, filename):
out = [] #why are you starting this with strings in it?
def checkMatch(line): #define the function to only take the line
#this would be the keyfunc for groupby in this example
return line.split(",",1)[0] #only split once, return the first element
with open(filename) as file:
for symb, lines in groupby(file,checkMatch):
#so here symb is the symbol at the start of each line of lines
if symb != wanted_symbol:
continue #skip this whole section if it has a different symbol
for line in csv.reader(lines):
out.append( [typ(val) for typ,val in zip(castings,line)] )
return out
I am having an issue with using the median function in numpy. The code used to work on a previous computer but when I tried to run it on my new machine, I got the error "cannot perform reduce with flexible type". In order to try to fix this, I attempted to use the map() function to make sure my list was a floating point and got this error message: could not convert string to float: .
Do some more attempts at debugging, it seems that my issue is with my splitting of the lines in my input file. The lines are of the form: 2456893.248202,4.490 and I want to split on the ",". However, when I print out the list for the second column of that line, I get
4
.
4
9
0
so it seems to somehow be splitting each character or something though I'm not sure how. The relevant section of code is below, I appreciate any thoughts or ideas and thanks in advance.
def curve_split(fn):
with open(fn) as f:
for line in f:
line = line.strip()
time,lc = line.split(",")
#debugging stuff
g=open('test.txt','w')
l1=map(lambda x:x+'\n',lc)
g.writelines(l1)
g.close()
#end debugging stuff
return time,lc
if __name__ == '__main__':
# place where I keep the lightcurve files from the image subtraction
dirname = '/home/kuehn/m4/kepler/subtraction/detrending'
files = glob.glob(dirname + '/*lc')
print(len(files))
# in order to create our lightcurve array, we need to know
# the length of one of our lightcurve files
lc0 = curve_split(files[0])
lcarr = np.zeros([len(files),len(lc0)])
# loop through every file
for i,fn in enumerate(files):
time,lc = curve_split(fn)
lc = map(float, lc)
# debugging
print(fn[5:58])
print(lc)
print(time)
# end debugging
lcm = lc/np.median(float(lc))
#lcm = ((lc[qual0]-np.median(lc[qual0]))/
# np.median(lc[qual0]))
lcarr[i] = lcm
print(fn,i,len(files))
I have a 5GB file of businesses and I'm trying to extract all the businesses that whose business type codes (SNACODE) start with the SNACODE corresponding to grocery stores. For example, SNACODEs for some businesses could be 42443013, 44511003, 44419041, 44512001, 44522004 and I want all businesses whose codes start with my list of grocery SNACODES codes = [4451,4452,447,772,45299,45291,45212]. In this case, I'd get the rows for 44511003, 44512001, and 44522004
Based on what I googled, the most efficient way to read in the file seemed to be one row at a time (if not the SQL route). I then used a for loop and checked if my SNACODE column started with any of my codes (which probably was a bad idea but the only way I could get to work).
I have no idea how many rows are in the file, but there are 84 columns. My computer was running for so long that I asked a friend who said it should only take 10-20 min to complete this task. My friend edited the code but I think he misunderstood what I was trying to do because his result returns nothing.
I am now trying to find a more efficient method than re-doing my 9.5 hours and having my laptop run for an unknown amount of time. The closest thing I've been able to find is most efficient way to find partial string matches in large file of strings (python), but it doesn't seem like what I was looking for.
Questions:
What's the best way to do this? How long should this take?
Is there any way that I can start where I stopped? (I have no idea how many rows of my 5gb file I read, but I have the last saved line of data--is there a fast/easy way to find the line corresponding to a unique ID in the file without having to read each line?)
This is what I tried -- in 9.5 hours it outputted a 72MB file (200k+ rows) of grocery stores
codes = [4451,4452,447,772,45299,45291,45212] #codes for grocery stores
for df in pd.read_csv('infogroup_bus_2010.csv',sep=',', chunksize=1):
data = np.asarray(df)
data = pd.DataFrame(data, columns = headers)
for code in codes:
if np.char.startswith(str(data["SNACODE"][0]), str(code)):
with open("grocery.csv", "a") as myfile:
data.to_csv(myfile, header = False)
print code
break #break code for loop if match
grocery.to_csv("grocery.csv", sep = '\t')
This is what my friend edited it to. I'm pretty sure the x = df[df.SNACODE.isin(codes)] is only matching perfect matches, and thus returning nothing.
codes = [4451,4452,447,772,45299,45291,45212]
matched = []
for df in pd.read_csv('infogroup_bus_2010.csv',sep=',', chunksize=1024*1024, dtype = str, low_memory=False):
x = df[df.SNACODE.isin(codes)]
if len(x):
matched.append(x)
print "Processed chunk and found {} matches".format(len(x))
output = pd.concat(matched, axis=0)
output.to_csv("grocery.csv", index = False)
Thanks!
To increase speed you could pre-build a single regexp matching the lines you need and the read the raw file lines (no csv parsing) and check them with the regexp...
codes = [4451,4452,447,772,45299,45291,45212]
col_number = 4 # Column number of SNACODE
expr = re.compile("[^,]*," * col_num +
"|".join(map(str, codes)) +
".*")
for L in open('infogroup_bus_2010.csv'):
if expr.match(L):
print L
Note that this is just a simple sketch as no escaping is considered... if the SNACODE column is not the first one and preceding fields may contain a comma you need a more sophisticated regexp like:
...
'([^"][^,]*,|"([^"]|"")*",)' * col_num +
...
that ignores commas inside double-quotes
You can probably make your pandas solution much faster:
codes = [4451, 4452, 447, 772, 45299, 45291, 45212]
codes = [str(code) for code in codes]
sna = pd.read_csv('infogroup_bus_2010.csv', usecols=['SNACODE'],
chunksize=int(1e6), dtype={'SNACODE': str})
with open('grocery.csv', 'w') as fout:
for chunk in sna:
for code in chunk['SNACODE']:
for target_code in codes:
if code.startswith(target_code):
fout.write('{}\n'.format(code))
Read only the needed column with usecols=['SNACODE']. You can adjust the chunk size with chunksize=int(1e6). Depending on your RAM you can likely make it much bigger.
I used to decode AIS messages with theis package (Python) https://github.com/schwehr/noaadata/tree/master/ais until I started getting a new format of the messages.
As you may know, AIS messages come in two types mostly. one part (one message) or two parts (multi message). Message#5 is always comes in two parts. example:
!AIVDM,2,1,1,A,55?MbV02;H;s<HtKR20EHE:address#hidden#Dn2222222216L961O5Gf0NSQEp6ClRp8,0*1C
!AIVDM,2,2,1,A,88888888880,2*25
I used to decode this just fine using the following piece of code:
nmeamsg = fields.split(',')
if nmeamsg[0] != '!AIVDM':
return
total = eval(nmeamsg[1])
part = eval(nmeamsg[2])
aismsg = nmeamsg[5]
nmeastring = string.join(nmeamsg[0:-1],',')
bv = binary.ais6tobitvec(aismsg)
msgnum = int(bv[0:6])
--
elif (total>1):
# Multi Slot Messages: 5,6,8,12,14,17,19,20?,21,24,26
global multimsg
if total==2:
if msgnum==5:
if nmeastring.count('!AIVDM')==2 and len(nmeamsg)==13: # make sure there are two parts concatenated together
aismsg = nmeamsg[5]+nmeamsg[11]
bv = binary.ais6tobitvec(aismsg)
msg5 = ais_msg_5.decode(bv)
print "message5 :",msg5
return msg5
Now I'm getting a new format of the messages:
!SAVDM,2,1,7,A,55#0hd01sq`pQ3W?O81L5#E:1=0U8U#000000016000006H0004m8523k#Dp,0*2A,1410825672
!SAVDM,2,2,7,A,4hC`2U#C`40,2*76,1410825672,1410825673
Note. the number at the last index is the time in epoch format
I tried to adjust my code to decode this new format. I succeed in decoding messages with one part. My problem is multi message type.
nmeamsg = fields.split(',')
if nmeamsg[0] != '!AIVDM' and nmeamsg[0] != '!SAVDM':
return
total = eval(nmeamsg[1])
part = eval(nmeamsg[2])
aismsg = nmeamsg[5]
nmeastring = string.join(nmeamsg[0:-1],',')
dbtimestring = time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(float(nmeamsg[7])))
bv = binary.ais6tobitvec(aismsg)
msgnum = int(bv[0:6])
Decoder can't bring the two lines as one. So decoding fails because message#5 should contain two strings not one. The error i get is in these lines:
if nmeastring.count('!SAVDM')==2 and len(nmeamsg)==13:
aismsg = nmeamsg[5]+nmeamsg[11]
Where len(nmeamsg) is always 8 (second line) and nmeastring.count('!SAVDM') is always 1
I hope I explained this clearly so someone can let me know what I'm missing here.
UPDATE
Okay I think I found the reason. I pass messages from file to script line by line:
for line in file:
i=i+1
try:
doais(line)
Where message#5 should be passed as two lines. Any idea on how can I accomplish that?
UPDATE
I did it by modifying the code a little bit:
for line in file:
i=i+1
try:
nmeamsg = line.split(',')
aismsg = nmeamsg[5]
bv = binary.ais6tobitvec(aismsg)
msgnum = int(bv[0:6])
print msgnum
if nmeamsg[0] != '!AIVDM' and nmeamsg[0] != '!SAVDM':
print "wrong format"
total = eval(nmeamsg[1])
if total == 1:
dbtimestring = time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(float(nmeamsg[8])))
doais(line,msgnum,dbtimestring,aismsg)
if total == 2: #Multi-line messages
lines= line+file.next()
nmeamsg = lines.split(',')
dbtimestring = time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(float(nmeamsg[15])))
aismsg = nmeamsg[5]+nmeamsg[12]
doais(lines,msgnum,dbtimestring,aismsg)
Be aware that noaadata is my old research code. libais is my production library thst is in use for NOAA's ERMA and WhaleAlert.
I usually make decoding a two pass process. First join multi-line messages. I refer to this as normalization (ais_normalize.py). You have several issues in this step. First the two component lines have different timestamps on the right of the second string. By the USCG old metadata standard, the last one matters. So my code will assume that these two lines are not related. Second, you don't have the required station id field.
Where are you getting the SA from in SAVDM? What device ("talker" in the NMEA vocab) is receiving these messages?
If you're in Ruby, I can recommend the NMEA and AIS decoder ruby gem that I wrote, available on github. It's based on the unofficial AIS spec at catb.org which is maintained by one of Kurt's colleagues.
It handles combining of multipart messages, reads from streams, and supports a large of NMEA and AIS messages. Decoding the 50 binary subtypes of AIS messages 6 and 8 is presently in development.
To handle the nonstandard lines you posted:
!SAVDM,2,1,7,A,55#0hd01sq`pQ3W?O81L5#E:1=0U8U#000000016000006H0004m8523k#Dp,0*2A,1410825672
!SAVDM,2,2,7,A,4hC`2U#C`40,2*76,1410825672,1410825673
It would be necessary to add a new parse rule that accepts fields after the checksum, but aside from that it should go smoothly. In other words, you'd copy the parser line here:
| BANG DATA CSUM { result = NMEAPlus::AISMessageFactory.create(val[0], val[1], val[2]) }
and have something like
| BANG DATA CSUM COMMA DATA { result = NMEAPlus::AISMessageFactory.create(val[0], val[1], val[2], val[4]) }
What do you do with those extra timestamp(s)? It almost looks like they've been appended by whatever software is doing the logging, rather than being part of the actual message.