I was experiencing some performance differences between python 2.7 and 3.5 when serializing pandas frames to CSV.
So did a quick search on google and found this benchmark:
https://gist.github.com/GitRay/4001b4962eb9f3e09a9d456ee5a30aae
And modified it a bit for my needs:
import pandas as pd
from time import time
import platform
def timeit(func, n=5):
start = time()
for i in range(n):
func()
end = time()
return (end - start) / n
def csvdumps(s):
s.to_csv('foo')
return 'foo'
def csvloads(fn):
return pd.read_csv(fn)
def hdfdumps(s):
s.to_hdf('foo', 'bar', mode='w')
return ('foo', 'bar')
def hdfloads(path):
return pd.read_hdf('foo', 'bar')
df = pd.DataFrame({'text': [str(i % 1000) for i in range(1000000)],
'numbers': range(1000000)})
keys = ['csv', 'hdfstore']
d = {'csv': [csvloads, csvdumps],
'hdfstore': [hdfloads, hdfdumps]}
result = dict()
for name, (loads, dumps) in d.items():
text = dumps(df.text)
numbers = dumps(df.numbers)
result[name] = {'text': {'dumps': timeit(lambda: dumps(df.text)),
'loads': timeit(lambda: loads(text))},
'numbers': {'dumps': timeit(lambda: dumps(df.numbers)),
'loads': timeit(lambda: loads(numbers))}}
########
# Plot #
########
# Much of this was taken from
# http://nbviewer.ipython.org/gist/mwaskom/886b4e5cb55fed35213d
# by Michael Waskom
import matplotlib.pyplot as plt
import seaborn as sns
sns.set(style="whitegrid", font_scale=1.3)
w, h = 7, 7
f, (left, right) = plt.subplots(nrows=1, ncols=2, sharex=True, figsize=(w*2, h), squeeze=True)
df = pd.DataFrame({'loads': [result[key]['text']['loads'] for key in keys],
'dumps': [result[key]['text']['dumps'] for key in keys],
'storage': keys})
df = pd.melt(df, "storage", value_name="duration", var_name="operation")
sns.barplot("duration", "storage", "operation", data=df, ax=left)
left.set(xlabel="Duration (s)", ylabel="")
sns.despine(bottom=True)
left.set_title('Cost to Serialize Text')
left.legend(loc="lower center", ncol=2, frameon=True, title="operation")
df = pd.DataFrame({'loads': [result[key]['numbers']['loads'] for key in keys],
'dumps': [result[key]['numbers']['dumps'] for key in keys],
'storage': keys})
df = pd.melt(df, "storage", value_name="duration", var_name="operation")
sns.barplot("duration", "storage", "operation", data=df, ax=right)
right.set(xlabel="Duration (s)", ylabel="")
sns.despine(bottom=True)
right.set_title('Cost to Serialize Numerical Data')
right.legend(loc="lower center", ncol=2, frameon=True, title="operation")
plt.savefig('serialize_py'+'.'.join(platform.python_version_tuple())+'.png')
As you can see in the results serializing in python 3 is much slower :
python 2.7 python 3.5 diff
load 0.3504s 0.329005s +06.50%
dump 1.2784s 3.333152s -61.65%
Does anybody know why?
Related
I was writing a script in python using google-cloud-datastore python module to upload data from my CSV to datastore. The script seems to work fine but There seems to be a problem that I'm stuck with. I see that the integer values from my CSV are being stored as Floating point number. Is it a default way of sending data to datastore or am I doing something wrong?
Here's my code:
import sys
import getopt
import pandas as pd
from google.cloud import datastore
def write_dict_chunks(data, SIZE=100):
log_count = 0
datastore_client = datastore.Client()
task_key = datastore_client.key(kind)
for i in xrange(0, len(data), SIZE):
entities = []
for each_entry in data[i : i+SIZE]:
nan_check = lambda v: v if str(v)!='nan' else None
string_check = lambda v: v.decode('utf-8') if isinstance(v, str) else v
write_row = {k: nan_check(string_check(v)) for k, v in each_entry.iteritems()}
entity = datastore.Entity(key=task_key)
entity.update(write_row)
entities.append(entity)
datastore_client.put_multi(entities)
log_count += len(entities)
print 'Wrote {} entities to datastore'.format(log_count)
try:
opts, args = getopt.getopt(sys.argv[1:], "ho:v", ["kind=", "filepath="])
if len(args) > 0:
for each in args:
print 'Unrecognized argument: '+each
sys.exit(2)
except getopt.GetoptError as err:
# print help information and exit:
print str(err) # will print something like "option -a not recognized"
print 'Usage: python parse_csv.py --kind=kind_name --filepath=path_to_csv'
kind = None
filepath = None
for option, argument in opts:
if option in '--kind':
kind = argument
elif option in '--filepath':
filepath = argument
df = pd.read_csv(filepath)
df = df.to_dict(orient='records')
write_dict_chunks(df)
I'm trying to run this code
from math import sqrt
import numpy as np
import warnings
from collections import Counter
import pandas as pd
import random
def k_nearest_neighbors(data,predict, k =3):
if len(data) >= k:
warnings.warn('K is set to a value less than total voting groups')
distances = []
for group in data:
for features in data[group]:
eucliden_distance = np.linalg.norm(np.array(features)-np.array(predict))
distances.append([eucliden_distance,group])
votes = [i[1] for i in sorted(distances)[:k]]
print(Counter(votes).most_common(1))
vote_result = Counter(votes).most_common(1)[0][0]
return vote_result
df = pd.read_csv('bc2.txt')
df.replace('?',-99999,inplace=True)
df.drop(['id'],1,inplace = True)
full_data = df.astype(float).values.tolist()
random.shuffle(full_data)
test_size = 0.2
train_set = {2:[],4:[]}
test_set = {2:[],4:[]}
train_data = full_data[:-int(test_size*len(full_data))]
test_data = full_data[-int(test_size*len(full_data)):]
for i in train_data:
train_set[i[-1]].append(i[:-1])
for i in train_data:
test_set[i[-1]].append(i[:-1])
correct = 0
total = 0
for group in test_set:
for data in test_set[group]:
vote = k_nearest_neighbors(train_set,data, k=5)
if group == vote:
correct += 1
total += 1
print ('Accuracy:',correct/total)
it comes out with this error msg
File "ml8.py", line 38, in <module>
train_set[i[-1]].append(i[:-1])
KeyError: 1.0
file m18.py is this above code file
below is the sample of txt file
id,clump_thickness,unif_cell_size,unif_cell_shape,marg_adhesion,single_epith_cell_size,bare_nuclei,bland_chrom,norm_nucleoli,mitoses,class
1000025,2,5,1,1,1,2,1,3,1,1
1002945,2,5,4,4,5,7,10,3,2,1
1015425,2,3,1,1,1,2,2,3,1,1
1016277,2,6,8,8,1,3,4,3,7,1
1017023,2,4,1,1,3,2,1,3,1,1
1017122,4,8,10,10,8,7,10,9,7,1
1018099,2,1,1,1,1,2,10,3,1,1
1018561,2,2,1,2,1,2,1,3,1,1
1033078,2,2,1,1,1,2,1,1,1,5
1033078,2,4,2,1,1,2,1,2,1,1
1035283,2,1,1,1,1,1,1,3,1,1
1036172,2,2,1,1,1,2,1,2,1,1
1041801,4,5,3,3,3,2,3,4,4,1
I'm using 2.7.11 version
Your train_set only contains keys 2 and 4, whereas your classes in that sample are 1 and 5.
Instead of using
train_set = {2:[],4:[]}
you might have better luck with defaultdict:
from collections import defaultdict
train_set = defaultdict(list)
This way a non-existent key will be initialized to a new empty list on first access.
I have read that improving feature selection will reduce the training time of my classifier and also improve accuracy but I am not sure how can I reduce the number of features. Should I count them and after select the first 3000 for example ?
This is my code :
def save_object(obj, filename):
with open(filename, 'wb') as output:
pickle.dump(obj,output,pickle.HIGHEST_PROTOCOL)
print "saved"
ujson.dumps({"output" : "obj"})
with open('neg5000.csv','rb') as f:
reader = csv.reader(f)
neg_tweets = list(reader)
print "list ready"
with open('pos5000.csv','rb') as f:
reader = csv.reader(f)
pos_tweets = list(reader)
print "list ready"
tweets = []
for (words, sentiment) in pos_tweets + neg_tweets:
words_filtered = [e.lower() for e in words.split() if len(e) >= 3]
tweets.append((words_filtered, sentiment))
def get_words_in_tweets(tweets):
all_words = []
for (words, sentiment) in tweets:
all_words.extend(words)
return all_words
def get_word_features(wordlist):
wordlist = nltk.FreqDist(wordlist)
word_features = list(wordlist.keys())[:3000]
#word_features = wordlist.keys()
return word_features
def extract_features(document):
document_words = set(document)
features = {}
for word in word_features:
features['contains(%s)' % word] = (word in document_words)
return features
#def extract_features(words):
# return dict([(word, True) for word in words])
word_features = get_word_features(get_words_in_tweets(tweets))
training_set = nltk.classify.apply_features(extract_features, tweets)
save_object(word_features, 'wordf.save')
print 'features done'
print datetime.datetime.now()
classifier = nltk.NaiveBayesClassifier.train(training_set)
print 'training done'
print datetime.datetime.now()
save_object(classifier, 'classifier.save')
tweet = 'I love this car'
print classifier.classify(extract_features(tweet.split()))
There's a number of ways to approach feature selection for the supervised classification problem (which is what Naive Bayes does). I suggest heading over to scikit-learn manual and just trying everything listed there, since the choice of particular method is dependends on the data you have.
The easiest way to do this is to switch to the scikit-learn implementation of Naive Bayes and the use a Pipeline to chain the feature selection and classifier training. See this tutorial for code examples.
Here's a version of your code using scikit-learn with SelectKBest feature selection:
import numpy as np
from sklearn.feature_extraction.text import CountVectorizer
from sklearn.feature_selection import SelectPercentile
from sklearn.naive_bayes import MultinomialNB
from sklearn.pipeline import Pipeline
def read_input(path):
with open(path) as handle:
lines = (line.rsplit(",", 1) for line in handle)
return [text for text, label in lines]
# Assuming each line in ``neg5000.csv`` and ``pos5000.csv`` is a
# UTF-8-encoded tweet.
neg_tweets = read_input("neg5000.csv")
pos_tweets = read_input("pos5000.csv")
X = np.append(neg_tweets, pos_tweets)
y = np.append(np.full(len(neg_tweets), -1, dtype=int),
np.full(len(pos_tweets), 1, dtype=int))
p = Pipeline([
("vectorizer", CountVectorizer()),
("selector", SelectPercentile(percentile=20)),
("nb", MultinomialNB())
])
p.fit(X, y)
print(p.predict(["I love this car"]))
I am trying to perform feature selection for logistic regression classifier. Originally there are 4 variables: name, location, gender, and label = ethnicity. The three variables, namely the name, give rise to tens of thousands of more "features", for example, name "John Snow" will give rise to 2-letter substrings like 'jo', 'oh', 'hn'... etc. The feature set undergoes DictVectorization.
I am trying to follow this tutorial (http://scikit-learn.org/stable/auto_examples/feature_selection/plot_feature_selection.html) but I am not sure if I am doing it right since the tutorial is using a small number of features while mine has tens of thousands after vectorization. And also the plt.show() shows a blank figure.
# coding=utf-8
import pandas as pd
from pandas import DataFrame, Series
import numpy as np
import re
import random
import time
from random import randint
import csv
import sys
reload(sys)
sys.setdefaultencoding('utf-8')
from sklearn import svm
from sklearn.metrics import classification_report
from sklearn.linear_model import LogisticRegression
from sklearn.svm import LinearSVC
from sklearn.tree import DecisionTreeClassifier
from sklearn.naive_bayes import MultinomialNB
from sklearn.feature_extraction import DictVectorizer
from sklearn.feature_selection import SelectPercentile, f_classif
from sklearn.metrics import confusion_matrix as sk_confusion_matrix
from sklearn.metrics import roc_curve, auc
import matplotlib.pyplot as plt
from sklearn.metrics import precision_recall_curve
# Assign X and y variables
X = df.raw_name.values
X2 = df.name.values
X3 = df.gender.values
X4 = df.location.values
y = df.ethnicity_scan.values
# Feature extraction functions
def feature_full_name(nameString):
try:
full_name = nameString
if len(full_name) > 1: # not accept name with only 1 character
return full_name
else: return '?'
except: return '?'
def feature_avg_wordLength(nameString):
try:
space = 0
for i in nameString:
if i == ' ':
space += 1
length = float(len(nameString) - space)
name_entity = float(space + 1)
avg = round(float(length/name_entity), 0)
return avg
except:
return 0
def feature_name_entity(nameString2):
space = 0
try:
for i in nameString2:
if i == ' ':
space += 1
return space+1
except: return 0
def feature_gender(genString):
try:
gender = genString
if len(gender) >= 1:
return gender
else: return '?'
except: return '?'
def feature_noNeighborLoc(locString):
try:
x = re.sub(r'^[^, ]*', '', locString) # remove everything before and include first ','
y = x[2:] # remove subsequent ',' and ' '
return y
except: return '?'
def list_to_dict(substring_list):
try:
substring_dict = {}
for i in substring_list:
substring_dict['substring='+str(i)] = True
return substring_dict
except: return '?'
# Transform format of X variables, and spit out a numpy array for all features
my_dict13 = [{'name-entity': feature_name_entity(feature_full_name(i))} for i in X2]
my_dict14 = [{'avg-length': feature_avg_wordLength(feature_full_name(i))} for i in X]
my_dict15 = [{'gender': feature_full_name(i)} for i in X3]
my_dict16 = [{'location': feature_noNeighborLoc(feature_full_name(i))} for i in X4]
my_dict17 = [{'dummy1': 1} for i in X]
my_dict18 = [{'dummy2': random.randint(0,2)} for i in X]
all_dict = []
for i in range(0, len(my_dict)):
temp_dict = dict(my_dict13[i].items() + my_dict14[i].items()
+ my_dict15[i].items() + my_dict16[i].items() + my_dict17[i].items() + my_dict18[i].items()
)
all_dict.append(temp_dict)
newX = dv.fit_transform(all_dict)
# Separate the training and testing data sets
half_cut = int(len(df)/2.0)*-1
X_train = newX[:half_cut]
X_test = newX[half_cut:]
y_train = y[:half_cut]
y_test = y[half_cut:]
# Fitting X and y into model, using training data
lr = LogisticRegression()
lr.fit(X_train, y_train)
dv = DictVectorizer()
# Feature selection
plt.figure(1)
plt.clf()
X_indices = np.arange(X_train.shape[-1])
selector = SelectPercentile(f_classif, percentile=10)
selector.fit(X_train, y_train)
scores = -np.log10(selector.pvalues_)
scores /= scores.max()
plt.bar(X_indices - .45, scores, width=.2,
label=r'Univariate score ($-Log(p_{value})$)', color='g')
plt.show()
Warning:
E:\Program Files Extra\Python27\lib\site-packages\sklearn\feature_selection\univariate_selection.py:111: UserWarning: Features [[0 0 0 ..., 0 0 0]] are constant.
It looks like the way you split your data into training and testing sets is not working:
# Separate the training and testing data sets
X_train = newX[:half_cut]
X_test = newX[half_cut:]
If you already use sklearn, it is much more convenient to use the builtin splitting routine for this:
X_train, X_test, y_train, y_test = cross_validation.train_test_split(X, y, test_size=0.5, random_state=0)
I updated the code and it now provides the graph, however after giving me the graph it produces the following error messages.
Warning (from warnings module):
File "C:\Python27\lib\site-packages\matplotlib\collections.py", line 590
if self._edgecolors == str('face'):
FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison
import urllib2
import time
import datetime
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.ticker as mticker
import matplotlib.dates as mdates
from matplotlib.finance import candlestick_ochl
import matplotlib
import pylab
matplotlib.rcParams.update({'font.size': 9})
def rsiFunc(prices, n=14):
deltas = np.diff(prices)
seed = deltas[:n+1]
up = seed[seed>=0].sum()/n
down = -seed[seed<0].sum()/n
rs = up/down
rsi = np.zeros_like(prices)
rsi[:n] = 100. - 100./(1.+rs)
for i in range(n, len(prices)):
delta = deltas[i-1] # cause the diff is 1 shorter
if delta>0:
upval = delta
downval = 0.
else:
upval = 0.
downval = -delta
up = (up*(n-1) + upval)/n
down = (down*(n-1) + downval)/n
rs = up/down
rsi[i] = 100. - 100./(1.+rs)
return rsi
def movingaverage(values,window):
weigths = np.repeat(1.0, window)/window
smas = np.convolve(values, weigths, 'valid')
return smas # as a numpy array
def ExpMovingAverage(values, window):
weights = np.exp(np.linspace(-1., 0., window))
weights /= weights.sum()
a = np.convolve(values, weights, mode='full')[:len(values)]
a[:window] = a[window]
return a
def computeMACD(x, slow=26, fast=12):
"""
compute the MACD (Moving Average Convergence/Divergence) using a fast and slow exponential moving avg'
return value is emaslow, emafast, macd which are len(x) arrays
"""
emaslow = ExpMovingAverage(x, slow)
emafast = ExpMovingAverage(x, fast)
return emaslow, emafast, emafast - emaslow
def graphData(stock,MA1,MA2):
'''
Use this to dynamically pull a stock:
'''
try:
print 'Currently Pulling',stock
print str(datetime.datetime.fromtimestamp(int(time.time())).strftime('%Y-%m-%d %H:%M:%S'))
#Keep in mind this is close high low open data from Yahoo
urlToVisit = 'http://chartapi.finance.yahoo.com/instrument/1.0/'+stock+'/chartdata;type=quote;range=10y/csv'
stockFile =[]
try:
sourceCode = urllib2.urlopen(urlToVisit).read()
splitSource = sourceCode.split('\n')
for eachLine in splitSource:
splitLine = eachLine.split(',')
if len(splitLine)==6:
if 'values' not in eachLine:
stockFile.append(eachLine)
except Exception, e:
print str(e), 'failed to organize pulled data.'
except Exception,e:
print str(e), 'failed to pull pricing data'
try:
date, closep, highp, lowp, openp, volume = np.loadtxt(stockFile,delimiter=',', unpack=True,
converters={ 0: mdates.strpdate2num('%Y%m%d')})
x = 0
y = len(date)
newAr = []
while x < y:
appendLine = date[x],openp[x],closep[x],highp[x],lowp[x],volume[x]
newAr.append(appendLine)
x+=1
Av1 = movingaverage(closep, MA1)
Av2 = movingaverage(closep, MA2)
SP = len(date[MA2-1:])
fig = plt.figure(facecolor='#07000d')
ax1 = plt.subplot2grid((6,4), (1,0), rowspan=4, colspan=4, axisbg='#07000d')
candlestick_ochl(ax1, newAr[-SP:], width=.6, colorup='#53c156', colordown='#ff1717')#width=.6, plot_day_summary_ohlc
Label1 = str(MA1)+' SMA'
Label2 = str(MA2)+' SMA'
ax1.plot(date[-SP:],Av1[-SP:],'#e1edf9',label=Label1, linewidth=1.5)
ax1.plot(date[-SP:],Av2[-SP:],'#4ee6fd',label=Label2, linewidth=1.5)
ax1.grid(True, color='w')
ax1.xaxis.set_major_locator(mticker.MaxNLocator(10))
ax1.xaxis.set_major_formatter(mdates.DateFormatter('%Y-%m-%d'))
ax1.yaxis.label.set_color("w")
ax1.spines['bottom'].set_color("#5998ff")
ax1.spines['top'].set_color("#5998ff")
ax1.spines['left'].set_color("#5998ff")
ax1.spines['right'].set_color("#5998ff")
ax1.tick_params(axis='y', colors='w')
plt.gca().yaxis.set_major_locator(mticker.MaxNLocator(prune='upper')) #gca()
ax1.tick_params(axis='x', colors='w')
plt.ylabel('Stock price and Volume')
maLeg = plt.legend(loc=9, ncol=2, prop={'size':7},
fancybox=True, borderaxespad=0.)
maLeg.get_frame().set_alpha(0.4)
textEd = plt.gca().get_legend().get_texts()#pylab.gca() changed to plt.gca()
plt.setp(textEd[0:5], color = 'w')#changed pylab.setp to plt.setp
volumeMin = 0
ax0 = plt.subplot2grid((6,4), (0,0), sharex=ax1, rowspan=1, colspan=4, axisbg='#07000d')
rsi = rsiFunc(closep)
rsiCol = '#c1f9f7'
posCol = '#386d13'
negCol = '#8f2020'
ax0.plot(date[-SP:], rsi[-SP:], rsiCol, linewidth=1.5)
ax0.axhline(70, color=negCol)
ax0.axhline(30, color=posCol)
ax0.fill_between(date[-SP:], rsi[-SP:], 70, where=(rsi[-SP:]>=70), facecolor=negCol, edgecolor=negCol, alpha=0.5)
ax0.fill_between(date[-SP:], rsi[-SP:], 30, where=(rsi[-SP:]<=30), facecolor=posCol, edgecolor=posCol, alpha=0.5)
ax0.set_yticks([30,70])
ax0.yaxis.label.set_color("w")
ax0.spines['bottom'].set_color("#5998ff")
ax0.spines['top'].set_color("#5998ff")
ax0.spines['left'].set_color("#5998ff")
ax0.spines['right'].set_color("#5998ff")
ax0.tick_params(axis='y', colors='w')
ax0.tick_params(axis='x', colors='w')
plt.ylabel('RSI')
ax1v = ax1.twinx()
ax1v.fill_between(date[-SP:],volumeMin, volume[-SP:], facecolor='#00ffe8', alpha=.4)
ax1v.axes.yaxis.set_ticklabels([])
ax1v.grid(False)
ax1v.set_ylim(0, 3*volume.max())
ax1v.spines['bottom'].set_color("#5998ff")
ax1v.spines['top'].set_color("#5998ff")
ax1v.spines['left'].set_color("#5998ff")
ax1v.spines['right'].set_color("#5998ff")
ax1v.tick_params(axis='x', colors='w')
ax1v.tick_params(axis='y', colors='w')
ax2 = plt.subplot2grid((6,4), (5,0), sharex=ax1, rowspan=1, colspan=4, axisbg='#07000d')
# START NEW INDICATOR CODE #
# END NEW INDICATOR CODE #
plt.gca().yaxis.set_major_locator(mticker.MaxNLocator(prune='upper'))
ax2.spines['bottom'].set_color("#5998ff")
ax2.spines['top'].set_color("#5998ff")
ax2.spines['left'].set_color("#5998ff")
ax2.spines['right'].set_color("#5998ff")
ax2.tick_params(axis='x', colors='w')
ax2.tick_params(axis='y', colors='w')
ax2.yaxis.set_major_locator(mticker.MaxNLocator(nbins=5, prune='upper'))
for label in ax2.xaxis.get_ticklabels():
label.set_rotation(45)
plt.suptitle(stock.upper(),color='w')
plt.setp(ax0.get_xticklabels(), visible=False)
plt.setp(ax1.get_xticklabels(), visible=False)
'''ax1.annotate('Big news!',(date[510],Av1[510]),
xytext=(0.8, 0.9), textcoords='axes fraction',
arrowprops=dict(facecolor='white', shrink=0.05),
fontsize=14, color = 'w',
horizontalalignment='right', verticalalignment='bottom')'''
plt.subplots_adjust(left=.09, bottom=.14, right=.94, top=.95, wspace=.20, hspace=0)
plt.show()
fig.savefig('example.png',facecolor=fig.get_facecolor())
except Exception,e:
print 'main loop',str(e)
while True:
stock = raw_input('Stock to plot: ')
graphData(stock,10,50)
Please look at the thread Violin plot: warning with matplotlib 1.4.3 and pyplot fill_between warning since upgrade of numpy to 1.10.10
It seems there is a bug in matplotlib 1.4.3 (which has only started causing that error since the upgrade to numpy 1.10). This is reportedly corrected in 1.5.0 (which should be released soon). Hope this helps.