Hey so currently I've been trying to convert the text containing column of a csv file into a dictionary. From here I would then like to create word embeddings (& potentially embeddings for subparts of words. ie. dictionary => dict - tion - nary) What would be the best way to go about this, what frameworks would work best? I have attached my current code and an example of one row of the database.
# First we must input the data, we can use pandas to do this.
import pandas as pd
# Our data does not have headers so we will fabricate ones ourself during the importing
data = pd.read_csv('agr_en_train.csv', header=None, names =['Unique_ID', 'Text', 'Aggression-level'])
# We can now check the data has loaded properly.
data
{0}{facebook_corpus_msr_1723796} {Well said sonu..you have courage to stand agai...} {OAG}
Please let me know if you require any other info the answer this question more adeptly. Additionally are there any recommended pre-created dictionaries. How would I utilise them, an answer or direction to any sources that would be helpful are greatly appreciated!
Related
I am working on creating a dataframe for classification tasks.
Since my data is coming from all kinds of different sources I am wondering what the best way to collect the data step by step would be.
I am starting off with a folder of files, and want to store their path and filename and add new data, such as their label, that I get from a txtfile that is saved somewhere else.
But what is the best way to do that?
I was thinking about a list of dictionary like
data = [{"path": path_to_file_1, "filename" : filename_1, "label" : label_1},
{"path": path_to_file_2, "filename" : filename_2, "label" : label_2},
{"path": path_to_file_3, "filename" : filename_3, "label" : label_3}]
and so on .
My idea was to iterate through my folder, collect the information via different functions that I wrote and create a dictionary for each of my files like so:
for filename in folder:
dict_filename={}
label=get_label(filename)
path=get_path(filename)
dict_filename["label"]=label
dict_filename["path"]=path
dict_filename["filename"]=filename
data.append(dict_filename)
with dict_filename being a dictionary that only contains the information of the file that I am looking at at the moment.
SO at the end I would get a list containing all the dictionaries that I created for all of my files.
My questions are:
Is this a way that makes sense or is there a different way that works better/easier/smoother?
If this works, what do I do to create a new dictionary in every loop (I suppose I need a different name for each dictionary so I just don't overwrite my first one with every loop)?
This might be something pretty basic as I am new to Python, but I am grateful for everyone that can help me out!
Thanks in advance!
The dictionary is the way to go on this one. However, there is a lot of redundancy that could be eliminated depending on the structure of your data.
For example, you can use one dictionary to store all the dataframes in this manner:
dfs[filename] = pd.DataFrame(path).rename(label)
This basically creates makes accessing the information much easier later on. In addition, you can use:
df = pd.concat(dfs, axis=1)
To combine all your dataframes in the end.
I have been working on Python for about 1.5yrs and looking for some direction. This is the first time I can't find what I need after doing a lot of searching and must be missing something- most likely searching the wrong terms.
Problem: I am working on an app that has many processes (Could be hundreds or even thousands). Each process may have a unique input and output data format - could be multiline strings, comma separated strings, excel or csv with or without varying headers and many others. I need something that will format the input correctly and handle the output based upon the process. New processes also need to be easily added/defined. I am open to whatever is the best approach, but my thoughts are to use a database that stores the template/data definition and use that to know the format given a process. However, I'm struggling to come up with exactly how, if this is really the best approach, but it needs to be a solution that is scalable. Any direction would be appreciated. Thank you.
A couple simple examples of data
Process 1 example data (multi line string with Header)
Input of
[ABC123, XYZ453, CDE987]
and the resulting data input below would be created:
Barcode
ABC123
XYZ453
CDE987
This code below works, but is not reusable for the example 2.
list = [ABC123, XYZ453, CDE987]
input = "Barcode /r/n"
for l in list:
input = input + l + '/r/n'
Process 2 example input template (comma separated with Header):
Barcode,Location,Param1,Param2
Item1,L1,11,A
Item1,L1,22,B
Item2,L1,33,C
Item2,L2,44,F
Item3,L2,55,B
Item3,L2,66,P
Process 2 example resulting input data (comma separated with Header):
Input of
{'Barcode':['ABC123', 'XYZ453', 'CDE987', 'FGH487', 'YTR123'], 'Location':['Shelf1', 'Shelf2']}
and using the template to create the input data below:
Barcode,Location,Param1,Param2
ABC123,Shelf1,11,A
ABC123,Shelf1,22,B
XYZ453,Shelf1,33,C
XYZ453,Shelf2,44,F
CDE987,Shelf2,55,B
CDE987,Shelf2,66,P
FGH487,Shelf1,11,A
FGH487,Shelf1,22,B
YTR123,Shelf1,33,C
YTR123,Shelf2,44,F
I know how to handle each process with hardcoded loop/dataframe merge, etc. Ive done some abstraction in other cases with dicts. However, how to define/store each format that vary so much and create reusable abstracted code is where I am stuck.
Maybe you can do the output of the functions as a tuple with the keys "datatype" and "output" for the actual output
This question already has answers here:
Is there a memory efficient and fast way to load big JSON files?
(11 answers)
Closed 3 years ago.
I would like to load one by one the items of my json file. The file could be up to 3gb so loading it in advance and looping over it is not an option.
My json file is basically a dictionary of key and value pairs (hundreds of pairs), and there is nothing I want to discard (ijson).
I just want to load one pair at a time to work with it. Is there anyway to do that?
So basically I found out in this answer how to do it in a much simple way:
https://stackoverflow.com/a/17326199/2933485
Using ijson, it looks like you can loop over the file without loadin it but opening the file and using ijson parse function over it, this is the example I found:
import ijson
for prefix, the_type, value in ijson.parse(open(json_file_name)):
print prefix, the_type, value
Why dont you populate a sqlite table with the data once and query the data using the record PK? See https://docs.python.org/3.7/library/sqlite3.html
OK, so json is a nested format, which means each repeating block (dict or list object) is surrounded by start and end characters. Normally, you read the entire file, and in doing so, can confirm the well-formed, structure and "closedness" of each object - in other words, it's verifiable that all objects are legally structured. When you load a json file into memory using the json library, part of that process is the validation.
If you want to do that for an extra large file - you have to forgo the normal library and roll your own, loading in a line (or chunk) at a time, and processing that under the assumption that validation will retrospectively succeed.
That's achievable (assuming you're able to put your faith in such an assumption) but it's probably something you'll have to write yourself.
One strategy might be to read a line at a time, splitting on the colon : character, with commas as record delimiters, which is a crude approximation of how key-value pairs are coded within json. Following this method, you're going to be able to process all but the first and final key-value pairs cleanly in sequence.
That just leaves you to write some special conditions for properly parsing the first and final records, which will come through garbled using this strategy.
Crudely then, call something like this (referencing the csv library) and treat the json like a massive, unusually formatted csv file.
import csv
with open('big.json', newline=',') as csv_json_franken_file:
jsonreader = csv.reader(csv_json_franken_file, delimiter=':', quotechar='"')
for row in jsonreader: # This bit reads in a "row" at a time, until finished
print(', '.join(row))
Then do some edge-case treatment of the first and last rows (more or less depending on the structure of your json) to repair the garbling caused by what is a fairly blatant hack. It's not clean, and it's not robust to changes in the content - but sometimes, you just have to play the hand you've been dealt.
To be honest, generating json files of 3GB in size is a little irresponsible, so if anyone comes asking, you've got that in your corner.
I have a CSV file with 100,000 rows.
Each row in column A is a sentence comprised of both chars and integers.
I want column B to contain only integers.
I want the new columns to be in the same CSV file.
How can I accomplish this?
If I'm understanding your question correctly, I would use .isdigit() to parse the data in column A. I'm frankly not sure what the format of column A is, so I don't know exactly what you would do with this (if you gave more information I could give a more specific answer). Your solution will likely come in a similar form to this:
def find(lines):
B = []
for line in lines:
numbers = [c for c in line if c.isdigit()]
current = int(''.join(numbers))
# current is the concatenation of all
# integers found in column A from left to right
B.append(current)
return B
Let me know if this makes sense or is even in the right track for your solution. Once again, without knowing what you're trying to do, and what A looks like, I'm not sure what your actual goals are.
EDIT
I'm not going to explain the csv stuff for you, mainly because there is a fantastic resource and library for it included in python here. If you have specific questions related to writing csv, definitely post them.
It sounds like you essentially want to pull int values out of column A then add them to a new column B. There are definitely many ways to solve this, but the general form of the problem is for each row you'll filter out the int, then you'll add the filtered int into the new column. I'll list a couple:
Regex: You could use a pattern such as [0-9]+ to pull the string out of A, then use int(whatever that output is) to cast to int, then store those values in B. I'm a sucker for a good regular expression and this one is fairly straight forward. Regexr is a great resource to learn about this and test your pattern.
Use an algorithm similar to above: The above algorithm worked before, but I've updated it slightly. Now that it's been updated it'll return an array of numbers correspondent to numbers in A from left to right. This is relatively sound, but it doesn't necessarily guarantee you have the right integer, given that if the title has an int in it, it'll mess some things up. It is likely one of the more clear ways of doing this, though.
csv data:
>c1,v1,c2,v2,Time
>13.9,412.1,29.7,177.2,14:42:01
>13.9,412.1,29.7,177.2,14:42:02
>13.9,412.1,29.7,177.2,14:42:03
>13.9,412.1,29.7,177.2,14:42:04
>13.9,412.1,29.7,177.2,14:42:05
>0.1,415.1,1.3,-0.9,14:42:06
>0.1,408.5,1.2,-0.9,14:42:07
>13.9,412.1,29.7,177.2,14:42:08
>0.1,413.4,1.3,-0.9,14:42:09
>0.1,413.8,1.3,-0.9,14:42:10
My current code that I have:
import pandas as pd
import csv
import datetime as dt
#Read .csv file, get timestamp and split it into date and time separately
Data = pd.read_csv('filedata.csv', parse_dates=['Time_Stamp'], infer_datetime_format=True)
Data['Date'] = Data.Time_Stamp.dt.date
Data['Time'] = Data.Time_Stamp.dt.time
#print (Data)
print (Data['Time_Stamp'])
Data['Time_Stamp'] = pd.to_datetime(Data['Time_Stamp'])
#Read timestamp within a certain range
mask = (Data['Time_Stamp'] > '2017-06-12 10:48:00') & (Data['Time_Stamp']<= '2017-06-12 11:48:00')
june13 = Data.loc[mask]
#print (june13)
What I'm trying to do is to read every 5 secs of data, and if 1 out of 5 secs of data of c1 is 10.0 and above, replace that value of c1 with 0.
I'm still new to python and I could not find examples for this. May I have some assistance as this problem is way beyond my python programming skills for now. Thank you!
I don't know the modules around csv files so my answer might look primitive, and I'm not quite sure what you are trying to accomplish here, but have you though of dealing with the file textually ?
From what I get, you want to read every c1, check the value and modify it.
To read and modify the file, you could do:
with open('filedata.csv', 'r+') as csv_file:
lines = csv_file.readlines()
# for each line, isolate data part and check - and modify, the first one if needed.
# I'm seriously not sure, you might have wanted to read only one out of five lines.
# For that, just do a while loop with an index, which increments through lines by 5.
for line in lines:
line = line.split(',') # split comma-separated-values
# Check condition and apply needed change.
if float(line[0]) >= 10:
line[0] = "0" # Directly as a string.
# Transform the list back into a single string.
",".join(line)
# Rewrite the file.
csv_file.seek(0)
csv_file.writelines(lines)
# Here you are ready to use the file just like you were already doing.
# Of course, the above code could be put in a function for known advantages.
(I don't have python here, so I couldn't test it and typos might be there.)
If you only need the dataframe without the file being modified:
Pretty much the same to be honest.
Instead of the file-writing at the end, you could do :
from io import StringIO # pandas needs stringIO instead of strings.
# Above code here, but without the last 6 lines.
Data = pd.read_csv(
StringIo("\n".join(lines)),
parse_dates=['Time_Stamp'],
infer_datetime_format=True
)
This should give you the Data you have, with changed values where needed.
Hope this wasn't completely off. Also, some people might find this approach horrible ; we have already coded working modules to do that kind of things, so why botter and dealing with the rough raw data ourselves ? Personally, I think that it's often much easier than learning all of the external modules I'll be using in my life if I don't try to understand how the text representation of files can be used. Your opinion might differ.
Also, this code might result in performances being lower, as we need to iterate through the text twice (pandas does it when reading). However, I don't think you'd get faster result by reading the csv like you already do, then iterate through data anyway to check condition. (You might win a cast per c1 checked value, but the difference is small and iterating through pandas dataframe might as well be slower than a list, depending on the state of their current optimisation.)
Of course, if you don't really need the pandas dataframe format, you could completely do it manually, it would take only a few more lines (or not, tbh) and shouldn't be slower, as the amount of iterations would be minimized : you could check conditions on data at the same time as you read it. It's getting late and I'm sure you can figure that out by yourself so I won't code it in my great editor (known as stackoverflow), ask if there's anything !