This commit is contained in:
Maciej Czajka 2022-04-02 14:23:36 +02:00
parent 7e9f103b63
commit 42f8e29583
3 changed files with 4 additions and 9 deletions

View File

@ -9,20 +9,15 @@ RUN pip3 install pandas
RUN pip3 install matplotlib RUN pip3 install matplotlib
RUN pip3 install sklearn RUN pip3 install sklearn
RUN apt-get install unzip RUN apt-get install unzip
#RUN mkdir root/.kaggle
#RUN touch root/.kaggle/kaggle.json
#RUN echo "{\"username\":\"maciejczajka\",\"key\":\"d8f044ad2b96d32372e22f9054c02bb1\"}" >> root/.kaggle/kaggle.json
#RUN chmod 666 root/.kaggle
#RUN chmod 666 root/.kaggle/kaggle.json
#ENV KAGGLE_USERNAME=maciejczajka
#ENV KAGGLE_KEY=d8f044ad2b96d32372e22f9054c02bb1
ENV CUTOFF=1000 ENV CUTOFF=1000
ARG KAGGLE_USERNAME ARG KAGGLE_USERNAME
ARG KAGGLE_KEY ARG KAGGLE_KEY
ENV KAGGLE_USERNAME=${KAGGLE_USERNAME} ENV KAGGLE_USERNAME=${KAGGLE_USERNAME}
ENV KAGGLE_KEY=${KAGGLE_KEY} ENV KAGGLE_KEY=${KAGGLE_KEY}
RUN mkdir /dane
WORKDIR /app WORKDIR /app
COPY download.sh . COPY download.sh .

View File

@ -152,7 +152,7 @@ def dataset_information(games):
# os.system("unzip top-video-games-19952021-metacritic.zip") # os.system("unzip top-video-games-19952021-metacritic.zip")
# wczytanie danych # wczytanie danych
games=pd.read_csv('/app/all_games1.csv') games=pd.read_csv('/dane/all_games1.csv')
# usunięcie 5 wyników "Stadia" # usunięcie 5 wyników "Stadia"
delete_stadia(games) delete_stadia(games)

View File

@ -8,7 +8,7 @@ dataset_operation() {
head -n $len2 all_games.csv.shuf > all_games.csv.test head -n $len2 all_games.csv.shuf > all_games.csv.test
head -n $len3 all_games.csv.shuf | tail -n $len2 > all_games.csv.dev head -n $len3 all_games.csv.shuf | tail -n $len2 > all_games.csv.dev
tail -n +$len4 all_games.csv.shuf > all_games.csv.train tail -n +$len4 all_games.csv.shuf > all_games.csv.train
cp all_games.csv.shuf /app/all_games1.csv cp all_games.csv /dane/all_games.csv
wc -l all_games.csv.* wc -l all_games.csv.*
} }