To make your training effective, you must practice the skills you will use during a game.
The most important skill you will use in a game is calculation. This is why time spent solving tactical problems is always well-spent, and will always help your performance in actual games. Fortunately, tactics are relatively easy to study on your own. You don’t even need a computer, just a book with diagrams. Some authorities recommend having a chess clock running while you solve tactical problems, but I do not go in for that degree of realism myself. Nor do I think it is necessary to set up a physical board: I think solving from the book diagram is fine. I do think it is important to do all the work in your head, without moving pieces around on a board to make it easier for yourself. That said, I am becoming less dogmatic in my old age, or perhaps a better word is lazier, and at times I do break down, set up a board, and move the pieces around when I am becoming frustrated. But I always feel guilty when I do it!
Endgames are less easy to practice. Here is what I do, and if anyone has a better way, I would like to hear it, because I am far from satisfied with my way.
First I read the explanations in the book about how to play a particular endgame position. Often I will ponder the diagram for a while first, and think about how I would approach the problem with my current knowledge.
Then, as part of reading the explanation in the book, I will usually set up the position on a board and play through the moves. If I have a question about why a certain move is not played, I may spend a bit of time trying to figure out the answer to my own question, moving the pieces around. My approach to learning endgames is much more hands-on, literally, than my approach to solving tactical problems.
When I feel I have a grasp of the ideas in a position, even if only a weak grasp, I will play out the position against Fritz, without the analysis window displayed.
Fritz will often not play like a human, which has plusses and minuses, mostly minuses in my experience. First the plusses. Sometimes Fritz will completely ignore the line of play recommended in the book, and stubbornly persist in playing the position its own way. I have found this to be useful, when Fritz finds a valid approach not explained in the book: then I have to come up with my own original way to deal with this new line of play. Another plus: when I make a tactical oversight, Fritz will punish me immediately.
Now for the minuses. The biggest minus is that Fritz, when defending a losing position, often fails to play the book moves. The book moves are usually tough for a human player to crack, which is why they are the book moves. Fritz, however, sees everything within its analytical horizon, and may suddenly pitch a piece or run away with the king like a chicken with its head cut off, obviously ceasing to resist, because the resulting checkmate may be a move or two further off than if Fritz had stuck with the book line. This is a big problem when you are trying to understand and overcome the sort of determined resistance a human player would put up against you.
Another minus: in positions requiring a more strategic approach, Fritz may not find the best plan unless you allocate enough time. True, Fritz will still typically find a good plan, and you may even learn more by having to face this different plan not explained in the book. But if you are trying to learn the specific plan explained in the book, this can be a minus. Your mileage may vary.
By the way, humans tend to devise multi-move plans based on logical concepts, while Fritz looks at every new position afresh and plays more tactically and opportunistically. If you do not play the moves Fritz thinks are best and therefore expects from you, Fritz has no qualms about turning on a dime and going in a totally new direction if the new direction seems more promising based on Fritz’s evaluation function. I am guessing this is why Fritz may sometimes vary from the book plan.
You see the essential problem I have found, when using Fritz to study the endgame: I don’t get the same kind of opposition I would from an intelligent human. As explained above, this situation is not without plusses, but it also has minuses.
I confess to being not very skillful with Fritz, let alone its partner-in-crime Chessbase, so I would hardly be surprised if others have come up with better training methods using computer-based tools than I have. There may even be existing software programs out there that address some of my concerns. I confess the limitations of my knowledge, and I am open to being instructed by others who are better informed.