Loading our dataset from disk is a fairly straightforward endeavor. As we previously mentioned, we will be slicing our data by date. To do this, we will need to convert the Unix epoch times in the dataset to more sliceable dates. This is easily accomplished with the pandas to_datetime() method, as shown in the following code:

def read_data():
    df = pd.read_csv("./data/bitcoin.csv")
    df["Time"] = pd.to_datetime(df.Timestamp, unit='s')
    df.index = df.Time
    df = df.drop(["Time", "Timestamp"], axis=1)
    return df