benbenbang

Quick'n easy gpg cheatsheet

• To create a key: gpg --full-generate-key generally you can select the defaults by gpg --gen-key, or you prefer a rsa4096 encrypted key like gpg --default-new-key-algo rsa4096 --gen-key

  NOTE: You "must" to associate your author/commiter email with your GPG key, if you didn't get prompted to enter the email, username, etc. You will need to

  • gpg --edit-key KeyID
  • you will see gpg>, type adduid and follow the instructions
  • type save

• To export a public key into file public.key:

benbenbang

# tensorflow full guide:
# https://developer.apple.com/metal/tensorflow-plugin/

# grpcio & h5py issue
export CFLAGS="-I/opt/homebrew/opt/openssl/include"
export LDFLAGS="-L/opt/homebrew/opt/openssl/lib"
export GRPC_PYTHON_BUILD_SYSTEM_OPENSSL=1
export GRPC_PYTHON_BUILD_SYSTEM_ZLIB=1

brew install hdf5

benbenbang

# Simple bash function to rewrite the wrong commiter & email
# > git-rewrite-commiter-email old@email.com new_username new@email.com

git-rewrite-commiter-email(){
    OLD_EMAIL="$1"
    NEW_USERNAME="$2"
    NEW_EMAIL="$3"
    if [[ -z $OLD_EMAIL || -z $NEW_USERNAME || -z NEW_EMAIL ]]; then
        echo "Positional args needed: \`OLD_EMAIL NEW_USERNAME NEW_EMAIL\`"
    else

benbenbang

# When creating virtulaenv with conda, if you're using PIP_REQUIRE_VIRTUALENV=true
# You might end up with you cannot install by pip even you already activated the virtualenv
# This patch is another approach to let your conda venv work with the PIP_REQUIRE_VIRTUALENV flag
# By patching your pip library with this file, and add another flag CONDA_ENVS in your bashrc
# For instance, export CONDA_ENVS="/Users/$USER/conda/envs"
# Then, the pip flag should not prevent you installing anything in your conda venv anymore.

python_path=$(which python)
python_version=$(echo "$(python -V 2>&1)" | grep -E -o '[0-9]{1,}\.[0-9]{1,}(\.[0-9]{1,})?' | grep -E -o '[0-9]{1,}.[0-9]{1,}')
pip_version=$(conda list | grep pip | grep -E -o '[0-9]{2,}\.[0-9]{1,}(\.[0-9]{1,})?' | grep -E -o '[0-9]{2,}.[0-9]{1,}')

benbenbang

import pandas as pd

data = {"name": ["A", "A", "B", "B", "B", "C", "C"], "expense": [2, 3.4, 5, 3, 2, 1, 7]}

for name, data in df.groupby("name"):
    # replace f-string if you're not using python 3.6 up
    locals()[f"df_{name}"] = data
    
"""
df_A

benbenbang

# Example
from sympy import *

x, y = symbols('x y')
f = x + 2*y + 1
g = Rational(3,2)*pi + exp(I*x) / (x**2 + y)

# coeff(x, n): As you can see, it is for finding the coefficient of "variables" in mathematical expressions.
# So if you want to get the constant, `coeff` is not the method you're looking for

benbenbang

lgbm_params = {
    "bagging_freq": 5,
    "bagging_fraction": 0.6,
    "bagging_seed": 123,
    "boost_from_average": "false",
    "boost": "gbdt",
    "feature_fraction": 0.3,
    "learning_rate": 0.005,
    "max_depth": 3,
    "metric": "rmse",

benbenbang

xgb_params = {
    "max_depth": 4,
    "eta": 0.01,
    "objective": "reg:squarederror",
    "eval_metric": ["rmse"],
    "booster": "gbtree",
    "verbosity": 0,
    "sample_type": "weighted",
    "max_delta_step": 4,
    "subsample": 0.5,

benbenbang

ngb = NGBRegressor(
    Base=default_tree_learner,
    Dist=Normal,
    Score=MLE,
    learning_rate=lr, # [0.009, 0.007]
    natural_gradient=True,
    minibatch_frac=0.2,
    n_estimators=est,  # [1000, 5000]
    verbose=False,
)

benbenbang

#####################################################
# Origine: Nanashi's solution # https://www.kaggle.com/jesucristo/1-house-prices-solution-top-1
# Rewrite and refine some processes
#######################################################

def make_data(test_size):
    dataframe = pd.read_csv(os.path.join("data", "train.csv"))
    
    ...