kernel_id
int64
24.2k
23.3M
prompt
stringlengths
8
1.85M
completetion
stringlengths
1
182k
comp_name
stringlengths
5
57
10,421,422
vocab = build_vocab(df['question_text']) print("Glove : ") oov_glove = check_coverage(vocab, embed_glove )<categorify>
titanic.SurnameFreq=titanic.TicketFreq
Titanic - Machine Learning from Disaster
10,421,422
def clean_numbers(x): x = re.sub('[0-9]{5,}', ' number ', x) x = re.sub('[0-9]{4}', ' number ', x) x = re.sub('[0-9]{3}', ' number ', x) x = re.sub('[0-9]{2}', ' number ', x) return x<feature_engineering>
titanic['Deck']=titanic['Cabin'].notnull().astype(str ).str[0] titanic['Deck'].value_counts()
Titanic - Machine Learning from Disaster
10,421,422
df['question_text'] = df['question_text'].apply(lambda x: clean_numbers(x))<compute_test_metric>
titanic=titanic.drop(['Cabin'],axis=1 )
Titanic - Machine Learning from Disaster
10,421,422
vocab = build_vocab(df['question_text']) print("Glove : ") oov_glove = check_coverage(vocab, embed_glove )<feature_engineering>
Titanic - Machine Learning from Disaster
10,421,422
train_df['treated_question'] = train_df['question_text'].apply(lambda x: x.lower()) train_df['treated_question'] = train_df['treated_question'].apply(lambda x: clean_contractions(x, contraction_mapping)) train_df['treated_question'] = train_df['treated_question'].apply(lambda x: clean_special_chars(x, punct, punct_map...
def FamilyGroup(family): a='' if family<=1: a='Single' elif family<=4: a='Small' else: a='Large' return a titanic['FamilyGroup']=titanic['Family'].map(FamilyGroup) titanic=titanic.drop(['Family'],axis=1 )
Titanic - Machine Learning from Disaster
10,421,422
test_df['treated_question'] = test_df['question_text'].apply(lambda x: x.lower()) test_df['treated_question'] = test_df['treated_question'].apply(lambda x: clean_contractions(x, contraction_mapping)) test_df['treated_question'] = test_df['treated_question'].apply(lambda x: clean_special_chars(x, punct, punct_mapping))...
def AgeGroup(age): a='' if age<=15: a='Child' elif age<=30: a='Young' elif age<=50: a='Adult' else: a='Old' return a titanic['AgeGroup']=titanic['Age'].map(AgeGroup) titanic=titanic.drop(['Age'],axis=1 )
Titanic - Machine Learning from Disaster
10,421,422
<split>
titanic=titanic.drop(['PassengerId','TicketFreq','Ticket','Fare','Title','Surname'], axis=1 )
Titanic - Machine Learning from Disaster
10,421,422
train, val = train_test_split(train_df, test_size=0.2, random_state=2) <prepare_x_and_y>
titanic_data=pd.get_dummies(titanic,columns=['Embarked','AgeGroup','Sex','Deck','FamilyGroup'] )
Titanic - Machine Learning from Disaster
10,421,422
xtrain = train['question_text'].fillna('_na_' ).values xval = val['question_text'].fillna('_na_' ).values xtest = test_df['question_text'].fillna('_na_' ).values<string_transform>
titanic_data.loc[891:1308]
Titanic - Machine Learning from Disaster
10,421,422
EMBED_SIZE = 300 MAX_FEATURES = 100000 MAXLEN = 60 tokenizer = Tokenizer(num_words=MAX_FEATURES) tokenizer.fit_on_texts(list(xtrain)) xtrain = tokenizer.texts_to_sequences(xtrain) xval = tokenizer.texts_to_sequences(xval) xtest = tokenizer.texts_to_sequences(xtest )<string_transform>
train_df = titanic_data.loc[0:890] train_df['Survived'] = train_results test_df = titanic_data.loc[891:1308]
Titanic - Machine Learning from Disaster
10,421,422
xtrain = pad_sequences(xtrain, maxlen=MAXLEN) xval = pad_sequences(xval, maxlen=MAXLEN) xtest = pad_sequences(xtest, maxlen=MAXLEN )<prepare_x_and_y>
X=train_df.drop(['Survived'],axis=1) X.head() y=train_df.Survived y.head()
Titanic - Machine Learning from Disaster
10,421,422
ytrain = train['target'].values yval = val['target'].values<statistical_test>
Titanic - Machine Learning from Disaster
10,421,422
def load_glove_matrix(word_index, embeddings_index): all_embs = np.stack(embeddings_index.values()) emb_mean, emb_std = all_embs.mean() , all_embs.std() EMBED_SIZE = all_embs.shape[1] nb_words = min(MAX_FEATURES, len(word_index)) embedding_matrix = np.random.normal(emb_mean, emb_std,(nb_words, EMBED_SIZE)) for word, i...
xgbr=XGBClassifier(n_estimators=2800, min_child_weight=0.1, learning_rate=0.002, max_depth=2, subsample=0.47, colsample_bytree=0.35, gamma=0.4, reg_lambda=0.4, random_state=42, n_jobs=-1,) xgbr.fit(X,y) predicts=xgbr.predict(test_df)
Titanic - Machine Learning from Disaster
10,421,422
np.random.seed(2) trn_idx = np.random.permutation(len(xtrain)) val_idx = np.random.permutation(len(xval)) xtrain = xtrain[trn_idx] ytrain = ytrain[trn_idx] xval = xval[val_idx] yval = yval[val_idx] embedding_matrix_glove = load_glove_matrix(tokenizer.word_index, embed_glove )<set_options>
submission = pd.DataFrame({ "PassengerId": test_data["PassengerId"], "Survived": predicts }) submission.Survived = submission.Survived.round().astype("int") submission.to_csv('titanic.csv', index=False) print("Submitted Successfully")
Titanic - Machine Learning from Disaster
1,646,791
class Attention(Layer): def __init__(self, step_dim, W_regularizer=None, b_regularizer=None, W_constraint=None, b_constraint=None, bias=True, **kwargs): self.supports_masking = True self.init = initializers.get('glorot_uniform') self.W_regularizer = regularizers.get(W_regularizer) self.b_regularizer = regularizers.ge...
%matplotlib inline warnings.filterwarnings('ignore')
Titanic - Machine Learning from Disaster
1,646,791
def f1(y_true, y_pred): def recall(y_true, y_pred): true_positives = K.sum(K.round(K.clip(y_true*y_pred, 0, 1))) possible_positives = K.sum(K.round(K.clip(y_true, 0, 1))) recall = true_positives/(possible_positives + K.epsilon()) return recall def precision(y_true, y_pred): true_positives = K.sum(K.round(K.clip(y_tr...
train = pd.read_csv('.. /input/train.csv') test = pd.read_csv('.. /input/test.csv') all_data = [train,test]
Titanic - Machine Learning from Disaster
1,646,791
def model_lstm_att(embedding_matrix): inp = Input(shape=(MAXLEN,)) x = Embedding(MAX_FEATURES, EMBED_SIZE, weights=[embedding_matrix], trainable=False )(inp) x = Bidirectional(CuDNNLSTM(64, return_sequences=True))(x) x = Bidirectional(CuDNNLSTM(32, return_sequences=True))(x) att = Attention(MAXLEN )(x) y = Dense(32...
cor_map(train.drop(['PassengerId'],axis=1))
Titanic - Machine Learning from Disaster
1,646,791
paragram = '.. /input/embeddings/paragram_300_sl999/paragram_300_sl999.txt' embedding_matrix_para = load_glove_matrix(tokenizer.word_index, load_embed(paragram))<compute_test_metric>
train[['Pclass','Survived']].groupby(['Pclass'], as_index=False ).mean().sort_values(by='Survived',ascending=False )
Titanic - Machine Learning from Disaster
1,646,791
embedding_matrix = np.mean([embedding_matrix_glove, embedding_matrix_para], axis=0 )<find_best_params>
train[['Sex','Survived']].groupby(['Sex'],as_index=False ).mean().sort_values(by='Survived',ascending=False )
Titanic - Machine Learning from Disaster
1,646,791
def train_pred(model, epochs=2): for e in range(epochs): model.fit(xtrain, ytrain, batch_size=512, epochs=3, validation_data=(xval, yval)) pred_val_y = model.predict([xval], batch_size=1024, verbose=1) best_thresh = 0.5 best_score = 0.0 for thresh in np.arange(0.1, 0.501, 0.01): thresh = np.round(thresh, 2) score = m...
train[['Embarked','Survived']].groupby(['Embarked'],as_index=False ).mean().sort_values(by='Survived',ascending=False )
Titanic - Machine Learning from Disaster
1,646,791
outputs = [] pred_val_y, pred_test_y, best_score = train_pred(model_lstm, epochs=2) outputs.append([pred_val_y, pred_test_y, best_score, 'model_lstm_att only Glove'] )<compute_test_metric>
guess_ages = np.zeros(( 3,9)) for dataset in all_data: dataset['ageFill']=dataset.Age.isnull().map({False:0,True:1}) med_all = dataset['Age'].median() for i in range(0,3): for j in range(0,9): guess_df=dataset[(dataset['Pclass']==i+1)&\ (dataset['SibSp']==j)]['Age'].dropna() age_guess=guess_df.median() try: guess_age...
Titanic - Machine Learning from Disaster
1,646,791
outputs.sort(key=lambda x: x[2]) weights = [i for i in range(1, len(outputs)+ 1)] weights = [float(i)/ sum(weights)for i in weights] pred_val_y = np.mean([outputs[i][0] for i in range(len(outputs)) ], axis = 0) thresholds = [] for thresh in np.arange(0.1, 0.501, 0.01): thresh = np.round(thresh, 2) res = metrics.f1_s...
freq_port = train.Embarked.dropna().mode() [0] train.Embarked = train.Embarked.fillna(freq_port) print(freq_port)
Titanic - Machine Learning from Disaster
1,646,791
print("Mejor limite:", best_thresh, "y puntuacion F1 ", thresholds[0][1] )<data_type_conversions>
test['Fare']=test.Fare.fillna(test.Fare.mean()) test.info()
Titanic - Machine Learning from Disaster
1,646,791
pred_test_y = np.mean([outputs[i][1] for i in range(len(outputs)) ], axis = 0) pred_test_y =(pred_test_y > best_thresh ).astype(int )<save_to_csv>
for dataset in all_data: dataset['Title'] = dataset.Name.str.extract('([A-Za-z]+)\.',expand=False) pd.crosstab(train['Title'],train['Sex'] )
Titanic - Machine Learning from Disaster
1,646,791
sub = pd.read_csv('.. /input/sample_submission.csv') out_df = pd.DataFrame({"qid":sub["qid"].values}) out_df['prediction'] = pred_test_y out_df.to_csv("submission.csv", index=False )<import_modules>
def cleanTicket(ticket): ticket = ticket.replace('.' , '') ticket = ticket.replace('/' , '') ticket = ticket.split() ticket = map(lambda t : t.strip() , ticket) ticket = list(filter(lambda t : not t.isdigit() , ticket)) if len(ticket)> 0: return ticket[0] else: return 'XXX' for dataset in all_data: dataset[ 'ticketP...
Titanic - Machine Learning from Disaster
1,646,791
ps = PorterStemmer() lc = LancasterStemmer() sb = SnowballStemmer("english") <set_options>
for dataset in all_data: title_mapping = {'Mr':1,'Rare':2,'Master':3,'Miss':4,'Mrs':5} dataset['Title'] = dataset['Title'].map(title_mapping) dataset['Title'] = dataset['Title'].fillna(0) train.head()
Titanic - Machine Learning from Disaster
1,646,791
gc.collect() K.clear_session()<normalization>
for dataset in all_data: dataset['Sex']=dataset.Sex.map({'male':0,'female':1}) train.head()
Titanic - Machine Learning from Disaster
1,646,791
class AttentionWeightedAverage(Layer): def __init__(self, return_attention=False, **kwargs): self.init = initializers.get('uniform') self.supports_masking = True self.return_attention = return_attention super(AttentionWeightedAverage, self ).__init__(** kwargs) def build(self, input_shape): self.input_spec = [InputSp...
for dataset in all_data: dataset['Embarked']=dataset.Embarked.map({'S':0,'Q':1,"C":2}) train.head()
Titanic - Machine Learning from Disaster
1,646,791
spell_model = gensim.models.KeyedVectors.load_word2vec_format('.. /input/embeddings/wiki-news-300d-1M/wiki-news-300d-1M.vec') words = spell_model.index2word w_rank = {} for i,word in enumerate(words): w_rank[word] = i WORDS = w_rank<set_options>
for dataset in all_data: dataset['AgeBand']=pd.cut(dataset['Age'],5,labels=[0,1,2,3,4]) dataset['AgeBand']=dataset.AgeBand.astype(int) train[['AgeBand','Survived']].groupby(['AgeBand'],as_index=False ).mean().sort_values(by='Survived',ascending=False )
Titanic - Machine Learning from Disaster
1,646,791
del spell_model, w_rank gc.collect()<categorify>
for dataset in all_data: dataset['cabinRec']=dataset.Cabin.isnull().map({False:0,True:1}) train.head()
Titanic - Machine Learning from Disaster
1,646,791
def words(text): return re.findall(r'\w+', text.lower()) def P(word): "Probability of `word`." return - WORDS.get(word, 0) def correction(word): "Most probable spelling correction for word." return max(candidates(word), key=P) def candidates(word): "Generate possible spelling corrections for word." return(known([wor...
for dataset in all_data: dataset['Fare'].fillna(dataset['Fare'].median() , inplace = True) dataset['FareBin'] = pd.qcut(dataset['Fare'], 5,labels=[1,2,3,4,5]) dataset['FareBin'] = dataset.FareBin.astype(int) train.head()
Titanic - Machine Learning from Disaster
1,646,791
def load_glove(word_dict, lemma_dict): EMBEDDING_FILE = '.. /input/embeddings/glove.840B.300d/glove.840B.300d.txt' def get_coefs(word,*arr): return word, np.asarray(arr, dtype='float32') embeddings_index = dict(get_coefs(*o.split(" ")) for o in open(EMBEDDING_FILE)) embed_size = 300 nb_words = len(word_dict)+1 embeddi...
Titanic - Machine Learning from Disaster
1,646,791
def load_fasttext(word_dict, lemma_dict): EMBEDDING_FILE = '.. /input/embeddings/wiki-news-300d-1M/wiki-news-300d-1M.vec' def get_coefs(word,*arr): return word, np.asarray(arr, dtype='float32') embeddings_index = dict(get_coefs(*o.split(" ")) for o in open(EMBEDDING_FILE)if len(o)>100) embed_size = 300 nb_words = len...
train_df=train.drop(['AgeBand','FareBin','Ticket','Cabin','Name','PassengerId','ticketPos'],axis=1) test_df=test.drop(['AgeBand','FareBin','Ticket','Cabin','Name','PassengerId','ticketPos'],axis=1) combine_df = [train_df,test_df] train_df.head()
Titanic - Machine Learning from Disaster
1,646,791
def load_para(word_dict, lemma_dict): EMBEDDING_FILE = '.. /input/embeddings/paragram_300_sl999/paragram_300_sl999.txt' def get_coefs(word,*arr): return word, np.asarray(arr, dtype='float32') embeddings_index = dict(get_coefs(*o.split(" ")) for o in open(EMBEDDING_FILE, encoding="utf8", errors='ignore')if len(o)>100) ...
for dataset in combine_df: dataset['Cabin_Class'] =(dataset.cabinRec+1)*dataset.Pclass
Titanic - Machine Learning from Disaster
1,646,791
def build_model(embedding_matrix, nb_words, embedding_size=300): inp = Input(shape=(max_length,)) x = Embedding(nb_words, embedding_size, weights=[embedding_matrix], trainable=False )(inp) x = SpatialDropout1D(0.3 )(x) x1 = Bidirectional(LSTM(256, return_sequences=True))(x) x2 = Bidirectional(GRU(128, return_sequenc...
train_df=train_df.drop('Sex',axis=1) test_df=test_df.drop('Sex',axis=1 )
Titanic - Machine Learning from Disaster
1,646,791
start_time = time.time() print("Loading data...") train = pd.read_csv(".. /input/train.csv" ).fillna(' ') test = pd.read_csv('.. /input/test.csv' ).fillna(' ') train_text = train['question_text'] test_text = test['question_text'] text_list = pd.concat([train_text, test_text]) y = train['target'].values num_train_da...
X_train_valid = train_df.drop('Survived',axis=1) y_train_valid = train_df['Survived'] X_test = test_df X_train,X_valid,y_train,y_valid=train_test_split(X_train_valid,y_train_valid,test_size=0.25,random_state=0) print(X_train.shape,X_test.shape,X_valid.shape )
Titanic - Machine Learning from Disaster
1,646,791
print("Start training...") start_time = time.time() model = build_model(embedding_matrix, nb_words, embedding_size) model.summary()<train_model>
gradb= GradientBoostingClassifier(learning_rate=0.01,random_state=0,n_estimators=2000,max_features=4) gradb.fit(X_train,y_train) print(gradb.score(X_valid,y_valid)) plot_model_var_imp(gradb,X_train,y_train )
Titanic - Machine Learning from Disaster
1,646,791
model.fit(train_word_sequences, y, batch_size=batch_size, epochs=num_epoch-1, verbose=2) pred_prob += 0.15*np.squeeze(model.predict(test_word_sequences, batch_size=batch_size, verbose=2)) model.fit(train_word_sequences, y, batch_size=batch_size, epochs=1, verbose=2) pred_prob += 0.35*np.squeeze(model.predict(test_wor...
clf = XGBClassifier(random_state=0,n_jobs=-1) cv_sets = ShuffleSplit(X_train.shape[0], n_iter =5, test_size = 0.20, random_state = 7) parameters = {'n_estimators':list(range(100,1000,100)) , 'learning_rate':[0.05,0.1,0.25,0.5,0.75], 'reg_lambda':[1,10,15,20,25]} acc_scorer=make_scorer(accuracy_score) grid_obj=GridSe...
Titanic - Machine Learning from Disaster
1,646,791
<train_model><EOS>
ids=test['PassengerId'] predictions = clf_best.predict(X_test) my_submission = pd.DataFrame({ 'PassengerId' : ids, 'Survived': predictions }) my_submission.to_csv('submission.csv', index=False)
Titanic - Machine Learning from Disaster
446,701
<SOS> metric: categorizationaccuracy Kaggle data source: titanic-machine-learning-from-disaster<load_from_csv>
import numpy as np import pandas as pd import matplotlib.pyplot as plt import seaborn as sns from sklearn.linear_model import LogisticRegression from sklearn.svm import SVC, LinearSVC from sklearn.ensemble import RandomForestClassifier from sklearn.neighbors import KNeighborsClassifier from sklearn.naive_bayes import G...
Titanic - Machine Learning from Disaster
446,701
train_df = pd.read_csv(".. /input/train.csv") test_df = pd.read_csv(".. /input/test.csv") print("Train shape : ",train_df.shape) print("Test shape : ",test_df.shape) train_df["question_text"] = train_df["question_text"].str.lower() test_df["question_text"] = test_df["question_text"].str.lower() def clean_text1(x): ...
titanic_df = pd.read_csv(".. /input/train.csv") test_df = pd.read_csv(".. /input/test.csv") titanic_df.head()
Titanic - Machine Learning from Disaster
446,701
np.random.seed(481945 )<import_modules>
def get_combined_data() : train = pd.read_csv('.. /input/train.csv') test = pd.read_csv('.. /input/test.csv') targets = train.Survived train.drop('Survived', 1, inplace=True) combined = train.append(test) combined.reset_index(inplace=True) combined.drop('index', inplace=True, axis=1) return combined
Titanic - Machine Learning from Disaster
446,701
from sklearn.metrics import pairwise<import_modules>
combined['Cabin'][combined.Cabin.isnull() ] = 'U0'
Titanic - Machine Learning from Disaster
446,701
from sklearn.feature_extraction.text import CountVectorizer from sklearn.model_selection import train_test_split import matplotlib.pyplot as plt from sklearn import metrics<categorify>
def get_titles() : global combined combined['Title'] = combined['Name'].map(lambda name:name.split(',')[1].split('.')[0].strip()) Title_Dictionary = { "Capt": "Officer", "Col": "Officer", "Major": "Officer", "Jonkheer": "Royalty", "Don": "Royalty", "Sir" : "Royalty", "Dr": "Officer", "Rev": "Officer", "the Countess":"...
Titanic - Machine Learning from Disaster
446,701
def summary(model, input_size, batch_size=-1, device="cuda", input_type=torch.float32): def register_hook(module): def hook(module, input, output): class_name = str(module.__class__ ).split(".")[-1].split("'")[0] module_idx = len(summary) m_key = "%s-%i" %(class_name, module_idx + 1) summary[m_key] = OrderedDict() su...
grouped_train = combined.head(891 ).groupby(['Sex','Pclass','Title']) grouped_median_train = grouped_train.median() grouped_test = combined.iloc[891:].groupby(['Sex','Pclass','Title']) grouped_median_test = grouped_test.median() grouped_median_train
Titanic - Machine Learning from Disaster
446,701
class CyclicLR(_LRScheduler): def __init__(self, optimizer, base_lr=1e-3, max_lr=6e-3, step_size_up=2000, step_size_down=None, mode='triangular', gamma=1., scale_fn=None, scale_mode='cycle', last_batch_idx=-1): if not isinstance(optimizer, Optimizer): raise TypeError('{} is not an Optimizer'.format( type(optimizer )...
def process_age() : global combined def fillAges(row, grouped_median): if row['Sex']=='female' and row['Pclass'] == 1: if row['Title'] == 'Miss': return grouped_median.loc['female', 1, 'Miss']['Age'] elif row['Title'] == 'Mrs': return grouped_median.loc['female', 1, 'Mrs']['Age'] elif row['Title'] == 'Officer': return ...
Titanic - Machine Learning from Disaster
446,701
quora_data = pd.read_csv('.. /input/train.csv' )<load_from_csv>
def process_names() : global combined combined.drop('Name',axis=1,inplace=True) titles_dummies = pd.get_dummies(combined['Title'],prefix='Title') combined = pd.concat([combined,titles_dummies],axis=1) combined.drop('Title',axis=1,inplace=True) status('names' )
Titanic - Machine Learning from Disaster
446,701
quora_test_data = pd.read_csv('.. /input/test.csv' )<define_variables>
def process_embarked() : global combined combined.head(891 ).Embarked.fillna('S', inplace=True) combined.iloc[891:].Embarked.fillna('S', inplace=True) embarked_dummies = pd.get_dummies(combined['Embarked'],prefix='Embarked') combined = pd.concat([combined,embarked_dummies],axis=1) combined.drop('Embarked',axis=1,in...
Titanic - Machine Learning from Disaster
446,701
for s in sample[sample.target == 1].question_text: print(s )<import_modules>
def process_cabin() : global combined combined.Cabin.fillna('U', inplace=True) combined['Cabin'] = combined['Cabin'].map(lambda c : c[0]) cabin_dummies = pd.get_dummies(combined['Cabin'], prefix='Cabin') combined = pd.concat([combined,cabin_dummies], axis=1) combined.drop('Cabin', axis=1, inplace=True) status('cab...
Titanic - Machine Learning from Disaster
446,701
from nltk.tokenize import TweetTokenizer<string_transform>
def process_sex() : global combined combined['Sex'] = combined['Sex'].map({'male':1,'female':0}) status('sex' )
Titanic - Machine Learning from Disaster
446,701
print(nltk.tokenize.word_tokenize("Don't spoil the movie or I'll kill you"))<string_transform>
def process_pclass() : global combined pclass_dummies = pd.get_dummies(combined['Pclass'], prefix="Pclass") combined = pd.concat([combined,pclass_dummies],axis=1) combined.drop('Pclass',axis=1,inplace=True) status('pclass' )
Titanic - Machine Learning from Disaster
446,701
print(TweetTokenizer().tokenize("Don't spoil the movie or I'll kill you"))<string_transform>
combined.drop('PassengerId', inplace=True, axis=1 )
Titanic - Machine Learning from Disaster
446,701
def tokenize(questions): tokenized_questions = [] for iteration, text in enumerate(questions): if iteration % 50000 == 0: print(iteration, "texts tokenized") tokenized_questions.append([t.lower() for t in TweetTokenizer().tokenize(text)]) return tokenized_questions<string_transform>
def process_ticket() : global combined def cleanTicket(ticket): ticket = ticket.replace('.','') ticket = ticket.replace('/','') ticket = ticket.split() ticket = map(lambda t : t.strip() , ticket) ticket = list(filter(lambda t : not t.isdigit() , ticket)) if len(ticket)> 0: return ticket[0] else: return 'XXX' combine...
Titanic - Machine Learning from Disaster
446,701
dev_tokens = tokenize(quora_data.question_text )<string_transform>
def process_family() : global combined combined['FamilySize'] = combined['Parch'] + combined['SibSp'] + 1 combined['Singleton'] = combined['FamilySize'].map(lambda s: 1 if s == 1 else 0) combined['SmallFamily'] = combined['FamilySize'].map(lambda s: 1 if 2<=s<=4 else 0) combined['LargeFamily'] = combined['FamilySize'...
Titanic - Machine Learning from Disaster
446,701
test_tokens = tokenize(quora_test_data.question_text )<import_modules>
pd.isnull(combined ).sum()
Titanic - Machine Learning from Disaster
446,701
import torchtext<import_modules>
def compute_score(clf, X, y, scoring='accuracy'): xval = cross_val_score(clf, X, y, cv = 5, scoring=scoring) return np.mean(xval )
Titanic - Machine Learning from Disaster
446,701
from collections import Counter import itertools<split>
def recover_train_test_target() : global combined train0 = pd.read_csv('.. /input/train.csv') targets = train0.Survived train = combined.head(891) test = combined.iloc[891:] return train, test, targets
Titanic - Machine Learning from Disaster
446,701
train_tokens, val_tokens, train_labels, val_labels = train_test_split(dev_tokens, quora_data.target, test_size=0.1 )<define_variables>
train.Fare.loc[50:]
Titanic - Machine Learning from Disaster
446,701
word_counts = Counter(itertools.chain(*train_tokens))<count_values>
clf = RandomForestClassifier(n_estimators=50, max_features='sqrt') clf = clf.fit(train,targets) print(clf.feature_importances_ )
Titanic - Machine Learning from Disaster
446,701
print(len(word_counts)) print(word_counts.most_common(10))<import_modules>
model = SelectFromModel(clf, prefit=True) train_reduced = model.transform(train) train_reduced.shape
Titanic - Machine Learning from Disaster
446,701
from gensim.models.keyedvectors import KeyedVectors<load_pretrained>
run_gs = False if run_gs: parameter_grid = { 'max_depth' : [4, 6, 8], 'n_estimators': [50, 10], 'max_features': ['sqrt', 'auto', 'log2'], 'min_samples_split': [1, 3, 10], 'min_samples_leaf': [1, 3, 10], 'bootstrap': [True, False], } forest = RandomForestClassifier() cross_validation = StratifiedKFold(targets, n_folds=5...
Titanic - Machine Learning from Disaster
446,701
gensim_vectors = KeyedVectors.load_word2vec_format('.. /input/embeddings/GoogleNews-vectors-negative300/GoogleNews-vectors-negative300.bin', binary=True )<feature_engineering>
compute_score(model, train, targets, scoring='accuracy' )
Titanic - Machine Learning from Disaster
446,701
def filter_vectors(gensim_vectors, words): result = {} for w in words: if w in gensim_vectors.vocab: result[w] = gensim_vectors[w].copy() return result<groupby>
output = model.predict(test ).astype(int) df_output = pd.DataFrame() aux = pd.read_csv('.. /input/test.csv') df_output['PassengerId'] = aux['PassengerId'] df_output['Survived'] = output df_output[['PassengerId','Survived']].to_csv('output.csv',index=False )
Titanic - Machine Learning from Disaster
11,512,965
filtered_vectors = filter_vectors(gensim_vectors, word_counts.keys() )<drop_column>
train = pd.read_csv('.. /input/titanic/train.csv') test_x = pd.read_csv('.. /input/titanic/test.csv') sub = pd.read_csv('.. /input/titanic/gender_submission.csv') df = pd.concat([train,test_x], sort = False) df.head()
Titanic - Machine Learning from Disaster
11,512,965
del gensim_vectors<count_values>
CabinFill = df[df['Cabin'].notnull() ] CabinNull = df[df['Cabin'].isnull() ] CabinFill['Cabin'] = CabinFill['Cabin'].astype(str ).str[0] df = pd.concat([CabinFill, CabinNull], sort = False ).sort_values(['PassengerId']) df['Sex'] = pd.Categorical(df.Sex ).codes df['Embarked'] = pd.Categorical(df.Embarked ).codes df['C...
Titanic - Machine Learning from Disaster
11,512,965
min_occurences = 14 filtered_counts = {w:c for w,c in word_counts.items() if c >= min_occurences} print(len(filtered_counts))<import_modules>
df.isnull().sum()
Titanic - Machine Learning from Disaster
11,512,965
class VocabLike: def __init__(self, itos, stoi): self.itos = itos self.stoi = stoi<import_modules>
df['Title'] = df['Name'].map(lambda name: name.split(',')[1].split('.')[0].strip()) df['Sur'] = df['Name'].map(lambda name: name.split(',')[0]) df['Title'].value_counts()
Titanic - Machine Learning from Disaster
11,512,965
from collections import defaultdict<feature_engineering>
titles_dummies = pd.get_dummies(df['Title'], prefix='Title') sur_dummies = pd.get_dummies(df['Sur'], prefix='Sur') df = pd.concat([df, titles_dummies, sur_dummies], axis=1) df.drop(['Title', 'Sur'], axis=1, inplace=True) df.head()
Titanic - Machine Learning from Disaster
11,512,965
specials = ['<unk>', '<pad>', '<eos>'] filtered_counts.update({w:0 for w in specials}) stoi = defaultdict(lambda:0) itos = [0] * len(filtered_counts) trainable_words = {w for w in filtered_counts.keys() if w not in specials and w not in filtered_vectors} pretrained_words = {w for w in filtered_counts.keys() if w not...
logloss, accuracy, y_preds = [],[],[] cv_train = np.zeros(( len(train),)) drop_cols = ['Name','PassengerId','Ticket','Survived','Fare','Cabin','Embarked', 'Age','SibSp','Parch'] train = df[:len(train)] train_x = train.drop(drop_cols, axis=1) train_y = train['Survived'] test = df[len(train):] test_x = test.drop(drop_co...
Titanic - Machine Learning from Disaster
11,512,965
def extract_vectors(vocab, vec_dict, offset, total, vec_size): vectors = np.zeros(( total, vec_size), dtype=np.float32) for i in range(total): word = vocab.itos[i + offset] assert word in vec_dict vectors[i] = vec_dict[word] return vectors<feature_engineering>
y_pred_cv =(cv_train > 0.5 ).astype(int) accuracy_score(train_y, y_pred_cv )
Titanic - Machine Learning from Disaster
11,512,965
np_vectors = extract_vectors(vocab, filtered_vectors, len(specials)+ len(trainable_words), len(pretrained_words), 300 )<statistical_test>
sub_y_lgb = sum(y_preds)/ len(y_preds) sub_y_lgb =(sub_y_lgb > 0.5 ).astype(int) sub['sub_y_lgb'] = sub_y_lgb
Titanic - Machine Learning from Disaster
11,512,965
def nearest_neighbors(vocab, embeddings, word, topn, use_offset=False): offset = len(specials)+ len(trainable_words)if use_offset else 0 assert word in vocab.stoi word_index = vocab.stoi[word] - offset sims = pairwise.cosine_similarity(embeddings[word_index].reshape(1,-1),embeddings ).ravel() indices = np.argsort(sims)...
accuracy, y_preds_LR = [],[] cv_train = np.zeros(( len(train),)) drop_cols = ['Name','PassengerId','Ticket','Survived','Fare','Cabin','Embarked', 'Age','SibSp','Parch'] train = df[:len(train)] train_x = train.drop(drop_cols, axis=1) train_y = train['Survived'] test = df[len(train):] test_x = test.drop(drop_cols, axis=...
Titanic - Machine Learning from Disaster
11,512,965
nearest_neighbors(vocab, np_vectors, 'we'll', 10, True )<set_options>
sub_y_LR = sum(y_preds_LR)/ len(y_preds_LR) sub_y_LR =(sub_y_LR > 0.5 ).astype(int) sub['sub_y_LR'] = sub_y_LR
Titanic - Machine Learning from Disaster
11,512,965
gc.collect()<categorify>
drop_cols = ['Name','PassengerId','Ticket','Survived'] train = df[:len(train)] X = train.drop(drop_cols, axis=1) y = train['Survived'] X_train, X_test, y_train, y_test = train_test_split( X, y, test_size=0.2, random_state=0) model = RandomForestClassifier() model.fit(X_train, y_train) y_pred = model.predict(X_test)...
Titanic - Machine Learning from Disaster
11,512,965
def to_word_indices(tokens, vocab, start_index, end_index): return [[start_index] + [vocab.stoi[w] for w in sent] + [end_index] for sent in tokens]<define_variables>
drop_cols = ['Name','PassengerId','Ticket','Survived'] test = df[len(train):] test_x = test.drop(drop_cols, axis=1) sub_y_RF = model.predict(test_x) sub['sub_y_RF'] = sub_y_RF
Titanic - Machine Learning from Disaster
11,512,965
class TokenToIdDataset(torch.utils.data.Dataset): def __init__(self, tokens, labels, vocab, max_size=-1, min_size=10, precompute=False, precomputed=False): self._start_index = vocab.stoi['<sos>'] self._end_index = vocab.stoi['<eos>'] self._pad_index = vocab.stoi['<pad>'] if precompute and not precomputed: tokens = to_w...
drop_cols = ['Name','PassengerId','Ticket','Survived','Fare','Cabin','Embarked', 'Age','SibSp','Parch'] x_NN = df.drop(drop_cols, axis=1) X_dummies_train = x_NN.iloc[0:890] X_dummies_test = x_NN.iloc[891:] Y = df.iloc[0:890]["Survived"]
Titanic - Machine Learning from Disaster
11,512,965
min_length=10 train_dataset = TokenToIdDataset(train_tokens,train_labels.values, vocab,min_size=min_length, precompute=True) val_dataset = TokenToIdDataset(val_tokens, val_labels.values, vocab,min_size=min_length, precompute=True )<set_options>
def create_neural_net(in_shape, lyrs=[4], act='relu', opt='Adam', dr=0.0): seed(37556) tf.random.set_seed(37556) model = Sequential() model.add(Dense(lyrs[0], input_dim=in_shape, activation=act)) for i in range(1,len(lyrs)) : model.add(Dense(lyrs[i], activation=act)) model.add(Dropout(dr)) model.add(Dense(1, activati...
Titanic - Machine Learning from Disaster
11,512,965
gc.collect()<import_modules>
single_net = create_neural_net(X_dummies_train.shape[1], lyrs =[4]) single_net.summary()
Titanic - Machine Learning from Disaster
11,512,965
import tqdm from tqdm import tqdm_notebook<load_pretrained>
training = single_net.fit(X_dummies_train, Y, epochs=100, batch_size=32, validation_split=0.25, verbose=0) val_acc = np.mean(training.history['val_accuracy']) print(" %s: %.2f%%" %('val_acc', val_acc*100))
Titanic - Machine Learning from Disaster
11,512,965
class BestModel: def __init__(self, model_path, optimizer_path, best_loss=10000): self.best_loss = best_loss self.model_path = model_path self.optimizer_path = optimizer_path def update(self, loss, model, optimizer=None): self.best_loss = loss torch.save(model.state_dict() , self.model_path) if optimizer: torch.save(o...
nn = KerasClassifier(build_fn=create_neural_net, in_shape = X_dummies_train.shape[1], lyrs=[12, 8, 4], epochs=50, dr=0.1, batch_size=1, verbose=0) nn.fit(X_dummies_train, Y) sub['sub_y_NN'] = nn.predict(X_dummies_test ).astype('int32' )
Titanic - Machine Learning from Disaster
11,512,965
<categorify><EOS>
model_cols = ['sub_y_lgb','sub_y_RF','sub_y_NN','sub_y_LR'] sub['Survived'] = np.sum(sub[model_cols], axis=1) sub['Survived'] =(sub['Survived'] >= 3 ).astype(int) sub.drop(model_cols, axis=1, inplace=True) sub.to_csv('sub_title_ensembled.csv', index=False) sub.head()
Titanic - Machine Learning from Disaster
3,861,474
<SOS> metric: categorizationaccuracy Kaggle data source: titanic-machine-learning-from-disaster<normalization>
import numpy as np import pandas as pd import os import torch import torch.nn as nn from torch.utils.data import Dataset, DataLoader import torch.nn.functional as F from sklearn.preprocessing import LabelEncoder
Titanic - Machine Learning from Disaster
3,861,474
class DotProductAttentionScoring(nn.Module): def __init__(self, scale): super().__init__() self.scale = np.sqrt(scale) def forward(self, query, keys): b_q, n_q, d_q = query.size() b_k, n_k, d_k = keys.size() assert b_q == b_k dot_products = torch.bmm(query, torch.transpose(keys, 1, 2)) / self.scale return dot_products...
train = pd.read_csv('.. /input/train.csv') test = pd.read_csv('.. /input/test.csv' )
Titanic - Machine Learning from Disaster
3,861,474
class TimeInvariant(nn.Module): def __init__(self, inner): super().__init__() self.inner = inner def forward(self, data): batch,time,dim = data.size() result = self.inner(data.view(batch * time, dim)) result = result.view(batch, time, -1) return result <categorify>
train = pd.read_csv('.. /input/train.csv') test = pd.read_csv('.. /input/test.csv' )
Titanic - Machine Learning from Disaster
3,861,474
class PositionalEncoding(nn.Module): "Implement the PE function." def __init__(self, d_model, max_len=5000): super(PositionalEncoding, self ).__init__() pe = torch.zeros(max_len, d_model,dtype=torch.float32) position = torch.arange(0., max_len ).unsqueeze(1) div_term = torch.exp(torch.arange(0., d_model, 2)* -(math.l...
all_df = pd.concat([train, test], sort=False )
Titanic - Machine Learning from Disaster
3,861,474
class SelfAttentionNet(nn.Module): def __init__(self, vocab, embeddings, num_trainable): super().__init__() self.pos_encoding = PositionalEncoding(50,120) self.num_trainable = num_trainable self.embedding = nn.Embedding(num_embeddings=len(vocab.itos), embedding_dim=300,padding_idx=vocab.stoi['<pad>']) self.embedding....
def preprocess(df, cat_cols): df = df.drop(['PassengerId', 'Name', 'Ticket', 'Cabin'], axis=1) for cat_col in cat_cols: if cat_col in ['Embarked']: df[cat_col] = LabelEncoder().fit_transform(df[cat_col].astype(str)) else: df[cat_col] = LabelEncoder().fit_transform(df[cat_col]) df = df.fillna(df.mean()) return df
Titanic - Machine Learning from Disaster
3,861,474
class Net(nn.Module): def __init__(self, vocab, embeddings, num_trainable, normalize=False): super().__init__() self.num_trainable = num_trainable self.embedding = nn.Embedding(num_embeddings=len(vocab.itos), embedding_dim=300,padding_idx=vocab.stoi['<pad>']) if normalize: embeddings = embeddings / np.linalg.norm(embe...
cat_cols = ['Pclass', 'Sex', 'SibSp', 'Parch', 'Embarked'] all_df = preprocess(all_df, cat_cols) all_df.head()
Titanic - Machine Learning from Disaster
3,861,474
num_scratch = len(specials)+ len(trainable_words )<choose_model_class>
class TabularDataset(Dataset): def __init__(self, df, categorical_columns, output_column=None): super().__init__() self.len = df.shape[0] self.categorical_columns = categorical_columns self.continous_columns = [col for col in df.columns if col not in self.categorical_columns + [output_column]] if self.continous_columns...
Titanic - Machine Learning from Disaster
3,861,474
net = Net(vocab, np_vectors, num_scratch ).cuda() best_model = BestModel('best_model', 'best_optimizer' )<choose_model_class>
train_ds = TabularDataset(train_df, cat_cols, 'Survived') train_dl = DataLoader(train_ds, 64, shuffle=True )
Titanic - Machine Learning from Disaster
3,861,474
criterion = nn.BCEWithLogitsLoss(pos_weight=torch.tensor([1.])).cuda() optimizer = torch.optim.Adam(net.parameters() , lr=0.0003 )<prepare_x_and_y>
class TitanicNet(nn.Module): def __init__(self, emb_dims, n_cont, lin_layer_sizes, output_size): super().__init__() self.emb_layers = nn.ModuleList([nn.Embedding(x, y)for x, y in emb_dims]) self.n_embs = sum([y for x, y in emb_dims]) self.n_cont = n_cont first_lin_layer = nn.Linear(self.n_embs + self.n_cont, lin_laye...
Titanic - Machine Learning from Disaster
3,861,474
def truncate_batch(batch): x,y = batch if x.shape[1] > 100: x = x[:,:100] return x,y<load_pretrained>
cat_dims = [int(all_df[col].nunique())for col in cat_cols] cat_dims
Titanic - Machine Learning from Disaster
3,861,474
train_loader = torch.utils.data.DataLoader(train_dataset, batch_size=256, collate_fn=lambda samples: truncate_batch(train_dataset.collate(samples)) , shuffle=True) val_loader = torch.utils.data.DataLoader(val_dataset, batch_size=256,collate_fn=lambda samples: truncate_batch(val_dataset.collate(samples))) <define_varia...
emb_dims = [(x, min(50,(x + 1)// 2)) for x in cat_dims] emb_dims
Titanic - Machine Learning from Disaster
3,861,474
print(128 * 112 * 112 * 300 * 4 / 1024 / 1024 )<train_model>
torch.manual_seed(2 )
Titanic - Machine Learning from Disaster
3,861,474
train_network(net, optimizer,criterion,train_loader,val_loader,16, 3,best_model, after_gradient=lambda epoch, network: network.zero_embedding_grad()) best_model.load(net, optimizer) <train_model>
model = TitanicNet(emb_dims, n_cont=2, lin_layer_sizes=[50, 100, 50], output_size=1) optimizer = torch.optim.Adam(model.parameters() , lr=0.003) no_of_epochs = 10 criterion = nn.BCELoss() for epoch in range(no_of_epochs): epoch_loss = 0 epoch_accuracy = 0 i = 0 for y, cont_x, cat_x in train_dl: preds = model(cont_x, ...
Titanic - Machine Learning from Disaster
3,861,474
train_network(net, optimizer,criterion,train_loader,val_loader,10, 3,best_model, after_gradient=lambda epoch, network: network.zero_embedding_grad()) best_model.load(net, optimizer) <load_pretrained>
test_df = all_df.tail(test.shape[0]) test_ds = TabularDataset(test_df, cat_cols, 'Survived') test_dl = DataLoader(test_ds, len(test_ds))
Titanic - Machine Learning from Disaster
3,861,474
best_model.load(net, optimizer )<create_dataframe>
with torch.no_grad() : for _, cont_x, cat_x in test_dl: preds = model(cont_x, cat_x) preds =(preds > 0.5 )
Titanic - Machine Learning from Disaster
3,861,474
best_model_with_fixed_embeddings = best_model.copy('best_model_fixed', 'best_optimizer_fixed' )<load_pretrained>
output_df = pd.DataFrame({'PassengerId':test['PassengerId'],'Survived':preds.flatten().numpy() } )
Titanic - Machine Learning from Disaster
3,861,474
<train_model><EOS>
output_df.to_csv('titanic_preds.csv', index=False )
Titanic - Machine Learning from Disaster
8,512,706
<SOS> metric: categorizationaccuracy Kaggle data source: titanic-machine-learning-from-disaster<load_pretrained>
pd.plotting.register_matplotlib_converters() %matplotlib inline %matplotlib inline %matplotlib inline plt.style.use('seaborn-whitegrid') warnings.filterwarnings('ignore' )
Titanic - Machine Learning from Disaster
8,512,706
best_model2 = best_model_with_fixed_embeddings.copy('best_model_low', 'best_optimizer_low') best_model2.load(net, optimizer) for g in optimizer.param_groups: g['lr'] = 0.00003 train_network(net, optimizer,criterion,train_loader,val_loader,10, 3,best_model2) best_model2.load(net, optimizer )<load_pretrained>
train=pd.read_csv("/kaggle/input/titanic/train.csv") X_test=pd.read_csv("/kaggle/input/titanic/test.csv") X_test.copy() train.info()
Titanic - Machine Learning from Disaster